Merge remote-tracking branch 'upstream/3.4' into merge-3.4

pull/14570/head
Alexander Alekhin 6 years ago
commit e21262deba
  1. 223
      doc/py_tutorials/py_video/py_lucas_kanade/py_lucas_kanade.markdown
  2. 183
      doc/py_tutorials/py_video/py_meanshift/py_meanshift.markdown
  3. 4
      doc/py_tutorials/py_video/py_table_of_contents_video.markdown
  4. 0
      doc/tutorials/video/meanshift/images/camshift_face.gif
  5. 0
      doc/tutorials/video/meanshift/images/camshift_result.jpg
  6. 0
      doc/tutorials/video/meanshift/images/meanshift_basics.jpg
  7. 0
      doc/tutorials/video/meanshift/images/meanshift_face.gif
  8. 0
      doc/tutorials/video/meanshift/images/meanshift_result.jpg
  9. 118
      doc/tutorials/video/meanshift/meanshift.markdown
  10. 0
      doc/tutorials/video/optical_flow/images/optical_flow_basic1.jpg
  11. 0
      doc/tutorials/video/optical_flow/images/opticalfb.jpg
  12. 0
      doc/tutorials/video/optical_flow/images/opticalflow_lk.jpg
  13. 156
      doc/tutorials/video/optical_flow/optical_flow.markdown
  14. 12
      doc/tutorials/video/table_of_content_video.markdown
  15. 7
      modules/calib3d/src/levmarq.cpp
  16. 2
      modules/calib3d/src/stereobm.cpp
  17. 123
      modules/core/include/opencv2/core/hal/intrin_avx.hpp
  18. 25
      modules/core/include/opencv2/core/hal/intrin_cpp.hpp
  19. 70
      modules/core/include/opencv2/core/hal/intrin_neon.hpp
  20. 131
      modules/core/include/opencv2/core/hal/intrin_sse.hpp
  21. 60
      modules/core/include/opencv2/core/hal/intrin_vsx.hpp
  22. 85
      modules/core/src/stat.simd.hpp
  23. 2
      modules/core/src/system.cpp
  24. 20
      modules/core/test/test_intrin_utils.hpp
  25. 6
      modules/dnn/src/dnn.cpp
  26. 49
      modules/dnn/src/layers/detection_output_layer.cpp
  27. 21
      modules/dnn/src/layers/proposal_layer.cpp
  28. 2
      modules/dnn/test/test_backends.cpp
  29. 4
      modules/dnn/test/test_caffe_importer.cpp
  30. 17
      modules/java/generator/android-21/java/org/opencv/android/JavaCamera2View.java
  31. 2
      samples/CMakeLists.example.in
  32. 48
      samples/CMakeLists.txt
  33. 6
      samples/cpp/CMakeLists.txt
  34. 2
      samples/cpp/example_cmake/CMakeLists.txt
  35. 4
      samples/cpp/tutorial_code/calib3d/real_time_pose_estimation/CMakeLists.txt
  36. 86
      samples/cpp/tutorial_code/video/meanshift/camshift.cpp
  37. 83
      samples/cpp/tutorial_code/video/meanshift/meanshift.cpp
  38. 101
      samples/cpp/tutorial_code/video/optical_flow/optical_flow.cpp
  39. 59
      samples/cpp/tutorial_code/video/optical_flow/optical_flow_dense.cpp
  40. 2
      samples/directx/CMakeLists.txt
  41. 2
      samples/dnn/CMakeLists.txt
  42. 201
      samples/dnn/object_detection.cpp
  43. 147
      samples/dnn/object_detection.py
  44. 6
      samples/gpu/CMakeLists.txt
  45. 2
      samples/opencl/CMakeLists.txt
  46. 4
      samples/opengl/CMakeLists.txt
  47. 2
      samples/openvx/CMakeLists.txt
  48. 50
      samples/python/tutorial_code/video/meanshift/camshift.py
  49. 49
      samples/python/tutorial_code/video/meanshift/meanshift.py
  50. 61
      samples/python/tutorial_code/video/optical_flow/optical_flow.py
  51. 23
      samples/python/tutorial_code/video/optical_flow/optical_flow_dense.py
  52. 31
      samples/samples_utils.cmake
  53. 2
      samples/tapi/CMakeLists.txt
  54. 2
      samples/va_intel/CMakeLists.txt

@ -1,225 +1,4 @@
Optical Flow {#tutorial_py_lucas_kanade}
============
Goal
----
In this chapter,
- We will understand the concepts of optical flow and its estimation using Lucas-Kanade
method.
- We will use functions like **cv.calcOpticalFlowPyrLK()** to track feature points in a
video.
Optical Flow
------------
Optical flow is the pattern of apparent motion of image objects between two consecutive frames
caused by the movemement of object or camera. It is 2D vector field where each vector is a
displacement vector showing the movement of points from first frame to second. Consider the image
below (Image Courtesy: [Wikipedia article on Optical
Flow](http://en.wikipedia.org/wiki/Optical_flow)).
![image](images/optical_flow_basic1.jpg)
It shows a ball moving in 5 consecutive frames. The arrow shows its displacement vector. Optical
flow has many applications in areas like :
- Structure from Motion
- Video Compression
- Video Stabilization ...
Optical flow works on several assumptions:
-# The pixel intensities of an object do not change between consecutive frames.
2. Neighbouring pixels have similar motion.
Consider a pixel \f$I(x,y,t)\f$ in first frame (Check a new dimension, time, is added here. Earlier we
were working with images only, so no need of time). It moves by distance \f$(dx,dy)\f$ in next frame
taken after \f$dt\f$ time. So since those pixels are the same and intensity does not change, we can say,
\f[I(x,y,t) = I(x+dx, y+dy, t+dt)\f]
Then take taylor series approximation of right-hand side, remove common terms and divide by \f$dt\f$ to
get the following equation:
\f[f_x u + f_y v + f_t = 0 \;\f]
where:
\f[f_x = \frac{\partial f}{\partial x} \; ; \; f_y = \frac{\partial f}{\partial y}\f]\f[u = \frac{dx}{dt} \; ; \; v = \frac{dy}{dt}\f]
Above equation is called Optical Flow equation. In it, we can find \f$f_x\f$ and \f$f_y\f$, they are image
gradients. Similarly \f$f_t\f$ is the gradient along time. But \f$(u,v)\f$ is unknown. We cannot solve this
one equation with two unknown variables. So several methods are provided to solve this problem and
one of them is Lucas-Kanade.
### Lucas-Kanade method
We have seen an assumption before, that all the neighbouring pixels will have similar motion.
Lucas-Kanade method takes a 3x3 patch around the point. So all the 9 points have the same motion. We
can find \f$(f_x, f_y, f_t)\f$ for these 9 points. So now our problem becomes solving 9 equations with
two unknown variables which is over-determined. A better solution is obtained with least square fit
method. Below is the final solution which is two equation-two unknown problem and solve to get the
solution.
\f[\begin{bmatrix} u \\ v \end{bmatrix} =
\begin{bmatrix}
\sum_{i}{f_{x_i}}^2 & \sum_{i}{f_{x_i} f_{y_i} } \\
\sum_{i}{f_{x_i} f_{y_i}} & \sum_{i}{f_{y_i}}^2
\end{bmatrix}^{-1}
\begin{bmatrix}
- \sum_{i}{f_{x_i} f_{t_i}} \\
- \sum_{i}{f_{y_i} f_{t_i}}
\end{bmatrix}\f]
( Check similarity of inverse matrix with Harris corner detector. It denotes that corners are better
points to be tracked.)
So from the user point of view, the idea is simple, we give some points to track, we receive the optical
flow vectors of those points. But again there are some problems. Until now, we were dealing with
small motions, so it fails when there is a large motion. To deal with this we use pyramids. When we go up in
the pyramid, small motions are removed and large motions become small motions. So by applying
Lucas-Kanade there, we get optical flow along with the scale.
Lucas-Kanade Optical Flow in OpenCV
-----------------------------------
OpenCV provides all these in a single function, **cv.calcOpticalFlowPyrLK()**. Here, we create a
simple application which tracks some points in a video. To decide the points, we use
**cv.goodFeaturesToTrack()**. We take the first frame, detect some Shi-Tomasi corner points in it,
then we iteratively track those points using Lucas-Kanade optical flow. For the function
**cv.calcOpticalFlowPyrLK()** we pass the previous frame, previous points and next frame. It
returns next points along with some status numbers which has a value of 1 if next point is found,
else zero. We iteratively pass these next points as previous points in next step. See the code
below:
@code{.py}
import numpy as np
import cv2 as cv
cap = cv.VideoCapture('slow.flv')
# params for ShiTomasi corner detection
feature_params = dict( maxCorners = 100,
qualityLevel = 0.3,
minDistance = 7,
blockSize = 7 )
# Parameters for lucas kanade optical flow
lk_params = dict( winSize = (15,15),
maxLevel = 2,
criteria = (cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 0.03))
# Create some random colors
color = np.random.randint(0,255,(100,3))
# Take first frame and find corners in it
ret, old_frame = cap.read()
old_gray = cv.cvtColor(old_frame, cv.COLOR_BGR2GRAY)
p0 = cv.goodFeaturesToTrack(old_gray, mask = None, **feature_params)
# Create a mask image for drawing purposes
mask = np.zeros_like(old_frame)
while(1):
ret,frame = cap.read()
frame_gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
# calculate optical flow
p1, st, err = cv.calcOpticalFlowPyrLK(old_gray, frame_gray, p0, None, **lk_params)
# Select good points
good_new = p1[st==1]
good_old = p0[st==1]
# draw the tracks
for i,(new,old) in enumerate(zip(good_new,good_old)):
a,b = new.ravel()
c,d = old.ravel()
mask = cv.line(mask, (a,b),(c,d), color[i].tolist(), 2)
frame = cv.circle(frame,(a,b),5,color[i].tolist(),-1)
img = cv.add(frame,mask)
cv.imshow('frame',img)
k = cv.waitKey(30) & 0xff
if k == 27:
break
# Now update the previous frame and previous points
old_gray = frame_gray.copy()
p0 = good_new.reshape(-1,1,2)
cv.destroyAllWindows()
cap.release()
@endcode
(This code doesn't check how correct are the next keypoints. So even if any feature point disappears
in image, there is a chance that optical flow finds the next point which may look close to it. So
actually for a robust tracking, corner points should be detected in particular intervals. OpenCV
samples comes up with such a sample which finds the feature points at every 5 frames. It also run a
backward-check of the optical flow points got to select only good ones. Check
samples/python/lk_track.py).
See the results we got:
![image](images/opticalflow_lk.jpg)
Dense Optical Flow in OpenCV
----------------------------
Lucas-Kanade method computes optical flow for a sparse feature set (in our example, corners detected
using Shi-Tomasi algorithm). OpenCV provides another algorithm to find the dense optical flow. It
computes the optical flow for all the points in the frame. It is based on Gunner Farneback's
algorithm which is explained in "Two-Frame Motion Estimation Based on Polynomial Expansion" by
Gunner Farneback in 2003.
Below sample shows how to find the dense optical flow using above algorithm. We get a 2-channel
array with optical flow vectors, \f$(u,v)\f$. We find their magnitude and direction. We color code the
result for better visualization. Direction corresponds to Hue value of the image. Magnitude
corresponds to Value plane. See the code below:
@code{.py}
import cv2 as cv
import numpy as np
cap = cv.VideoCapture("vtest.avi")
ret, frame1 = cap.read()
prvs = cv.cvtColor(frame1,cv.COLOR_BGR2GRAY)
hsv = np.zeros_like(frame1)
hsv[...,1] = 255
while(1):
ret, frame2 = cap.read()
next = cv.cvtColor(frame2,cv.COLOR_BGR2GRAY)
flow = cv.calcOpticalFlowFarneback(prvs,next, None, 0.5, 3, 15, 3, 5, 1.2, 0)
mag, ang = cv.cartToPolar(flow[...,0], flow[...,1])
hsv[...,0] = ang*180/np.pi/2
hsv[...,2] = cv.normalize(mag,None,0,255,cv.NORM_MINMAX)
bgr = cv.cvtColor(hsv,cv.COLOR_HSV2BGR)
cv.imshow('frame2',bgr)
k = cv.waitKey(30) & 0xff
if k == 27:
break
elif k == ord('s'):
cv.imwrite('opticalfb.png',frame2)
cv.imwrite('opticalhsv.png',bgr)
prvs = next
cap.release()
cv.destroyAllWindows()
@endcode
See the result below:
![image](images/opticalfb.jpg)
OpenCV comes with a more advanced sample on dense optical flow, please see
samples/python/opt_flow.py.
Additional Resources
--------------------
Exercises
---------
-# Check the code in samples/python/lk_track.py. Try to understand the code.
2. Check the code in samples/python/opt_flow.py. Try to understand the code.
Tutorial content has been moved: @ref tutorial_optical_flow

@ -1,185 +1,4 @@
Meanshift and Camshift {#tutorial_py_meanshift}
======================
Goal
----
In this chapter,
- We will learn about Meanshift and Camshift algorithms to find and track objects in videos.
Meanshift
---------
The intuition behind the meanshift is simple. Consider you have a set of points. (It can be a pixel
distribution like histogram backprojection). You are given a small window ( may be a circle) and you
have to move that window to the area of maximum pixel density (or maximum number of points). It is
illustrated in the simple image given below:
![image](images/meanshift_basics.jpg)
The initial window is shown in blue circle with the name "C1". Its original center is marked in blue
rectangle, named "C1_o". But if you find the centroid of the points inside that window, you will
get the point "C1_r" (marked in small blue circle) which is the real centroid of window. Surely
they don't match. So move your window such that circle of the new window matches with previous
centroid. Again find the new centroid. Most probably, it won't match. So move it again, and continue
the iterations such that center of window and its centroid falls on the same location (or with a
small desired error). So finally what you obtain is a window with maximum pixel distribution. It is
marked with green circle, named "C2". As you can see in image, it has maximum number of points. The
whole process is demonstrated on a static image below:
![image](images/meanshift_face.gif)
So we normally pass the histogram backprojected image and initial target location. When the object
moves, obviously the movement is reflected in histogram backprojected image. As a result, meanshift
algorithm moves our window to the new location with maximum density.
### Meanshift in OpenCV
To use meanshift in OpenCV, first we need to setup the target, find its histogram so that we can
backproject the target on each frame for calculation of meanshift. We also need to provide initial
location of window. For histogram, only Hue is considered here. Also, to avoid false values due to
low light, low light values are discarded using **cv.inRange()** function.
@code{.py}
import numpy as np
import cv2 as cv
cap = cv.VideoCapture('slow.flv')
# take first frame of the video
ret,frame = cap.read()
# setup initial location of window
r,h,c,w = 250,90,400,125 # simply hardcoded the values
track_window = (c,r,w,h)
# set up the ROI for tracking
roi = frame[r:r+h, c:c+w]
hsv_roi = cv.cvtColor(roi, cv.COLOR_BGR2HSV)
mask = cv.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv.normalize(roi_hist,roi_hist,0,255,cv.NORM_MINMAX)
# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 1 )
while(1):
ret ,frame = cap.read()
if ret == True:
hsv = cv.cvtColor(frame, cv.COLOR_BGR2HSV)
dst = cv.calcBackProject([hsv],[0],roi_hist,[0,180],1)
# apply meanshift to get the new location
ret, track_window = cv.meanShift(dst, track_window, term_crit)
# Draw it on image
x,y,w,h = track_window
img2 = cv.rectangle(frame, (x,y), (x+w,y+h), 255,2)
cv.imshow('img2',img2)
k = cv.waitKey(60) & 0xff
if k == 27:
break
else:
cv.imwrite(chr(k)+".jpg",img2)
else:
break
cv.destroyAllWindows()
cap.release()
@endcode
Three frames in a video I used is given below:
![image](images/meanshift_result.jpg)
Camshift
--------
Did you closely watch the last result? There is a problem. Our window always has the same size when
car is farther away and it is very close to camera. That is not good. We need to adapt the window
size with size and rotation of the target. Once again, the solution came from "OpenCV Labs" and it
is called CAMshift (Continuously Adaptive Meanshift) published by Gary Bradsky in his paper
"Computer Vision Face Tracking for Use in a Perceptual User Interface" in 1998.
It applies meanshift first. Once meanshift converges, it updates the size of the window as,
\f$s = 2 \times \sqrt{\frac{M_{00}}{256}}\f$. It also calculates the orientation of best fitting ellipse
to it. Again it applies the meanshift with new scaled search window and previous window location.
The process is continued until required accuracy is met.
![image](images/camshift_face.gif)
### Camshift in OpenCV
It is almost same as meanshift, but it returns a rotated rectangle (that is our result) and box
parameters (used to be passed as search window in next iteration). See the code below:
@code{.py}
import numpy as np
import cv2 as cv
cap = cv.VideoCapture('slow.flv')
# take first frame of the video
ret,frame = cap.read()
# setup initial location of window
r,h,c,w = 250,90,400,125 # simply hardcoded the values
track_window = (c,r,w,h)
# set up the ROI for tracking
roi = frame[r:r+h, c:c+w]
hsv_roi = cv.cvtColor(roi, cv.COLOR_BGR2HSV)
mask = cv.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv.normalize(roi_hist,roi_hist,0,255,cv.NORM_MINMAX)
# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 1 )
while(1):
ret ,frame = cap.read()
if ret == True:
hsv = cv.cvtColor(frame, cv.COLOR_BGR2HSV)
dst = cv.calcBackProject([hsv],[0],roi_hist,[0,180],1)
# apply meanshift to get the new location
ret, track_window = cv.CamShift(dst, track_window, term_crit)
# Draw it on image
pts = cv.boxPoints(ret)
pts = np.int0(pts)
img2 = cv.polylines(frame,[pts],True, 255,2)
cv.imshow('img2',img2)
k = cv.waitKey(60) & 0xff
if k == 27:
break
else:
cv.imwrite(chr(k)+".jpg",img2)
else:
break
cv.destroyAllWindows()
cap.release()
@endcode
Three frames of the result is shown below:
![image](images/camshift_result.jpg)
Additional Resources
--------------------
-# French Wikipedia page on [Camshift](http://fr.wikipedia.org/wiki/Camshift). (The two animations
are taken from here)
2. Bradski, G.R., "Real time face and object tracking as a component of a perceptual user
interface," Applications of Computer Vision, 1998. WACV '98. Proceedings., Fourth IEEE Workshop
on , vol., no., pp.214,219, 19-21 Oct 1998
Exercises
---------
-# OpenCV comes with a Python sample on interactive demo of camshift. Use it, hack it, understand
it.
Tutorial content has been moved: @ref tutorial_meanshift

@ -1,13 +1,13 @@
Video Analysis {#tutorial_py_table_of_contents_video}
==============
- @subpage tutorial_py_meanshift
- @ref tutorial_meanshift
We have already seen
an example of color-based tracking. It is simpler. This time, we see significantly better
algorithms like "Meanshift", and its upgraded version, "Camshift" to find and track them.
- @subpage tutorial_py_lucas_kanade
- @ref tutorial_optical_flow
Now let's discuss an important concept, "Optical Flow", which is related to videos and has many applications.

@ -0,0 +1,118 @@
Meanshift and Camshift {#tutorial_meanshift}
======================
Goal
----
In this chapter,
- We will learn about the Meanshift and Camshift algorithms to track objects in videos.
Meanshift
---------
The intuition behind the meanshift is simple. Consider you have a set of points. (It can be a pixel
distribution like histogram backprojection). You are given a small window (may be a circle) and you
have to move that window to the area of maximum pixel density (or maximum number of points). It is
illustrated in the simple image given below:
![image](images/meanshift_basics.jpg)
The initial window is shown in blue circle with the name "C1". Its original center is marked in blue
rectangle, named "C1_o". But if you find the centroid of the points inside that window, you will
get the point "C1_r" (marked in small blue circle) which is the real centroid of the window. Surely
they don't match. So move your window such that the circle of the new window matches with the previous
centroid. Again find the new centroid. Most probably, it won't match. So move it again, and continue
the iterations such that the center of window and its centroid falls on the same location (or within a
small desired error). So finally what you obtain is a window with maximum pixel distribution. It is
marked with a green circle, named "C2". As you can see in the image, it has maximum number of points. The
whole process is demonstrated on a static image below:
![image](images/meanshift_face.gif)
So we normally pass the histogram backprojected image and initial target location. When the object
moves, obviously the movement is reflected in the histogram backprojected image. As a result, the meanshift
algorithm moves our window to the new location with maximum density.
### Meanshift in OpenCV
To use meanshift in OpenCV, first we need to setup the target, find its histogram so that we can
backproject the target on each frame for calculation of meanshift. We also need to provide an initial
location of window. For histogram, only Hue is considered here. Also, to avoid false values due to
low light, low light values are discarded using **cv.inRange()** function.
@add_toggle_cpp
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/cpp/tutorial_code/video/meanshift/meanshift.cpp)
- **Code at glance:**
@include samples/cpp/tutorial_code/video/meanshift/meanshift.cpp
@end_toggle
@add_toggle_python
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/python/tutorial_code/video/meanshift/meanshift.py)
- **Code at glance:**
@include samples/python/tutorial_code/video/meanshift/meanshift.py
@end_toggle
Three frames in a video I used is given below:
![image](images/meanshift_result.jpg)
Camshift
--------
Did you closely watch the last result? There is a problem. Our window always has the same size whether
the car is very far or very close to the camera. That is not good. We need to adapt the window
size with size and rotation of the target. Once again, the solution came from "OpenCV Labs" and it
is called CAMshift (Continuously Adaptive Meanshift) published by Gary Bradsky in his paper
"Computer Vision Face Tracking for Use in a Perceptual User Interface" in 1998 @cite Bradski98 .
It applies meanshift first. Once meanshift converges, it updates the size of the window as,
\f$s = 2 \times \sqrt{\frac{M_{00}}{256}}\f$. It also calculates the orientation of the best fitting ellipse
to it. Again it applies the meanshift with new scaled search window and previous window location.
The process continues until the required accuracy is met.
![image](images/camshift_face.gif)
### Camshift in OpenCV
It is similar to meanshift, but returns a rotated rectangle (that is our result) and box
parameters (used to be passed as search window in next iteration). See the code below:
@add_toggle_cpp
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/cpp/tutorial_code/video/meanshift/camshift.cpp)
- **Code at glance:**
@include samples/cpp/tutorial_code/video/meanshift/camshift.cpp
@end_toggle
@add_toggle_python
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/python/tutorial_code/video/meanshift/camshift.py)
- **Code at glance:**
@include samples/python/tutorial_code/video/meanshift/camshift.py
@end_toggle
Three frames of the result is shown below:
![image](images/camshift_result.jpg)
Additional Resources
--------------------
-# French Wikipedia page on [Camshift](http://fr.wikipedia.org/wiki/Camshift). (The two animations
are taken from there)
2. Bradski, G.R., "Real time face and object tracking as a component of a perceptual user
interface," Applications of Computer Vision, 1998. WACV '98. Proceedings., Fourth IEEE Workshop
on , vol., no., pp.214,219, 19-21 Oct 1998
Exercises
---------
-# OpenCV comes with a Python [sample](https://github.com/opencv/opencv/blob/master/samples/python/camshift.py) for an interactive demo of camshift. Use it, hack it, understand
it.

@ -0,0 +1,156 @@
Optical Flow {#tutorial_optical_flow}
============
Goal
----
In this chapter,
- We will understand the concepts of optical flow and its estimation using Lucas-Kanade
method.
- We will use functions like **cv.calcOpticalFlowPyrLK()** to track feature points in a
video.
- We will create a dense optical flow field using the **cv.calcOpticalFlowFarneback()** method.
Optical Flow
------------
Optical flow is the pattern of apparent motion of image objects between two consecutive frames
caused by the movemement of object or camera. It is 2D vector field where each vector is a
displacement vector showing the movement of points from first frame to second. Consider the image
below (Image Courtesy: [Wikipedia article on Optical Flow](http://en.wikipedia.org/wiki/Optical_flow)).
![image](images/optical_flow_basic1.jpg)
It shows a ball moving in 5 consecutive frames. The arrow shows its displacement vector. Optical
flow has many applications in areas like :
- Structure from Motion
- Video Compression
- Video Stabilization ...
Optical flow works on several assumptions:
-# The pixel intensities of an object do not change between consecutive frames.
2. Neighbouring pixels have similar motion.
Consider a pixel \f$I(x,y,t)\f$ in first frame (Check a new dimension, time, is added here. Earlier we
were working with images only, so no need of time). It moves by distance \f$(dx,dy)\f$ in next frame
taken after \f$dt\f$ time. So since those pixels are the same and intensity does not change, we can say,
\f[I(x,y,t) = I(x+dx, y+dy, t+dt)\f]
Then take taylor series approximation of right-hand side, remove common terms and divide by \f$dt\f$ to
get the following equation:
\f[f_x u + f_y v + f_t = 0 \;\f]
where:
\f[f_x = \frac{\partial f}{\partial x} \; ; \; f_y = \frac{\partial f}{\partial y}\f]\f[u = \frac{dx}{dt} \; ; \; v = \frac{dy}{dt}\f]
Above equation is called Optical Flow equation. In it, we can find \f$f_x\f$ and \f$f_y\f$, they are image
gradients. Similarly \f$f_t\f$ is the gradient along time. But \f$(u,v)\f$ is unknown. We cannot solve this
one equation with two unknown variables. So several methods are provided to solve this problem and
one of them is Lucas-Kanade.
### Lucas-Kanade method
We have seen an assumption before, that all the neighbouring pixels will have similar motion.
Lucas-Kanade method takes a 3x3 patch around the point. So all the 9 points have the same motion. We
can find \f$(f_x, f_y, f_t)\f$ for these 9 points. So now our problem becomes solving 9 equations with
two unknown variables which is over-determined. A better solution is obtained with least square fit
method. Below is the final solution which is two equation-two unknown problem and solve to get the
solution.
\f[\begin{bmatrix} u \\ v \end{bmatrix} =
\begin{bmatrix}
\sum_{i}{f_{x_i}}^2 & \sum_{i}{f_{x_i} f_{y_i} } \\
\sum_{i}{f_{x_i} f_{y_i}} & \sum_{i}{f_{y_i}}^2
\end{bmatrix}^{-1}
\begin{bmatrix}
- \sum_{i}{f_{x_i} f_{t_i}} \\
- \sum_{i}{f_{y_i} f_{t_i}}
\end{bmatrix}\f]
( Check similarity of inverse matrix with Harris corner detector. It denotes that corners are better
points to be tracked.)
So from the user point of view, the idea is simple, we give some points to track, we receive the optical
flow vectors of those points. But again there are some problems. Until now, we were dealing with
small motions, so it fails when there is a large motion. To deal with this we use pyramids. When we go up in
the pyramid, small motions are removed and large motions become small motions. So by applying
Lucas-Kanade there, we get optical flow along with the scale.
Lucas-Kanade Optical Flow in OpenCV
-----------------------------------
OpenCV provides all these in a single function, **cv.calcOpticalFlowPyrLK()**. Here, we create a
simple application which tracks some points in a video. To decide the points, we use
**cv.goodFeaturesToTrack()**. We take the first frame, detect some Shi-Tomasi corner points in it,
then we iteratively track those points using Lucas-Kanade optical flow. For the function
**cv.calcOpticalFlowPyrLK()** we pass the previous frame, previous points and next frame. It
returns next points along with some status numbers which has a value of 1 if next point is found,
else zero. We iteratively pass these next points as previous points in next step. See the code
below:
@add_toggle_cpp
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/cpp/tutorial_code/video/optical_flow/optical_flow.cpp)
- **Code at glance:**
@include samples/cpp/tutorial_code/video/optical_flow/optical_flow.cpp
@end_toggle
@add_toggle_python
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/python/tutorial_code/video/optical_flow/optical_flow.py)
- **Code at glance:**
@include samples/python/tutorial_code/video/optical_flow/optical_flow.py
@end_toggle
(This code doesn't check how correct are the next keypoints. So even if any feature point disappears
in image, there is a chance that optical flow finds the next point which may look close to it. So
actually for a robust tracking, corner points should be detected in particular intervals. OpenCV
samples comes up with such a sample which finds the feature points at every 5 frames. It also run a
backward-check of the optical flow points got to select only good ones. Check
samples/python/lk_track.py).
See the results we got:
![image](images/opticalflow_lk.jpg)
Dense Optical Flow in OpenCV
----------------------------
Lucas-Kanade method computes optical flow for a sparse feature set (in our example, corners detected
using Shi-Tomasi algorithm). OpenCV provides another algorithm to find the dense optical flow. It
computes the optical flow for all the points in the frame. It is based on Gunner Farneback's
algorithm which is explained in "Two-Frame Motion Estimation Based on Polynomial Expansion" by
Gunner Farneback in 2003.
Below sample shows how to find the dense optical flow using above algorithm. We get a 2-channel
array with optical flow vectors, \f$(u,v)\f$. We find their magnitude and direction. We color code the
result for better visualization. Direction corresponds to Hue value of the image. Magnitude
corresponds to Value plane. See the code below:
@add_toggle_cpp
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/cpp/tutorial_code/video/optical_flow/optical_flow_dense.cpp)
- **Code at glance:**
@include samples/cpp/tutorial_code/video/optical_flow/optical_flow_dense.cpp
@end_toggle
@add_toggle_python
- **Downloadable code**: Click
[here](https://github.com/opencv/opencv/tree/master/samples/python/tutorial_code/video/optical_flow/optical_flow_dense.py)
- **Code at glance:**
@include samples/python/tutorial_code/video/optical_flow/optical_flow_dense.py
@end_toggle
See the result below:
![image](images/opticalfb.jpg)

@ -14,3 +14,15 @@ tracking and foreground extractions.
We will learn how to extract foreground masks from both videos and sequences of images and
to show them.
- @subpage tutorial_meanshift
*Languages:* C++, Python
Learn how to use the Meanshift and Camshift algorithms to track objects in videos.
- @subpage tutorial_optical_flow
*Languages:* C++, Python
We will learn how to use optical flow methods to track sparse features or to create a dense representation.

@ -80,11 +80,8 @@ namespace cv
class LMSolverImpl CV_FINAL : public LMSolver
{
public:
LMSolverImpl() : maxIters(100) { init(); }
LMSolverImpl(const Ptr<LMSolver::Callback>& _cb, int _maxIters) : cb(_cb), epsx(FLT_EPSILON), epsf(FLT_EPSILON), maxIters(_maxIters) { init(); }
LMSolverImpl(const Ptr<LMSolver::Callback>& _cb, int _maxIters, double _eps) : cb(_cb), epsx(_eps), epsf(_eps), maxIters(_maxIters) { init(); }
void init()
LMSolverImpl(const Ptr<LMSolver::Callback>& _cb, int _maxIters, double _eps = FLT_EPSILON)
: cb(_cb), epsx(_eps), epsf(_eps), maxIters(_maxIters)
{
printInterval = 0;
}

@ -1130,7 +1130,7 @@ public:
CV_Error( Error::StsOutOfRange, "SADWindowSize must be odd, be within 5..255 and be not larger than image width or height" );
if( params.numDisparities <= 0 || params.numDisparities % 16 != 0 )
CV_Error( Error::StsOutOfRange, "numDisparities must be positive and divisble by 16" );
CV_Error( Error::StsOutOfRange, "numDisparities must be positive and divisible by 16" );
if( params.textureThreshold < 0 )
CV_Error( Error::StsOutOfRange, "texture threshold must be non-negative" );

@ -1015,6 +1015,34 @@ OPENCV_HAL_IMPL_AVX_ROTATE_CAST(v_rotate_right, v_float64x4, _mm256_castsi256_pd
////////// Reduce and mask /////////
/** Reduce **/
inline unsigned v_reduce_sum(const v_uint8x32& a)
{
__m256i half = _mm256_sad_epu8(a.val, _mm256_setzero_si256());
__m128i quarter = _mm_add_epi32(_v256_extract_low(half), _v256_extract_high(half));
return (unsigned)_mm_cvtsi128_si32(_mm_add_epi32(quarter, _mm_unpackhi_epi64(quarter, quarter)));
}
inline int v_reduce_sum(const v_int8x32& a)
{
__m256i half = _mm256_sad_epu8(_mm256_xor_si256(a.val, _mm256_set1_epi8((schar)-128)), _mm256_setzero_si256());
__m128i quarter = _mm_add_epi32(_v256_extract_low(half), _v256_extract_high(half));
return (unsigned)_mm_cvtsi128_si32(_mm_add_epi32(quarter, _mm_unpackhi_epi64(quarter, quarter))) - 4096;
}
#define OPENCV_HAL_IMPL_AVX_REDUCE_32(_Tpvec, sctype, func, intrin) \
inline sctype v_reduce_##func(const _Tpvec& a) \
{ \
__m128i val = intrin(_v256_extract_low(a.val), _v256_extract_high(a.val)); \
val = intrin(val, _mm_srli_si128(val,8)); \
val = intrin(val, _mm_srli_si128(val,4)); \
val = intrin(val, _mm_srli_si128(val,2)); \
val = intrin(val, _mm_srli_si128(val,1)); \
return (sctype)_mm_cvtsi128_si32(val); \
}
OPENCV_HAL_IMPL_AVX_REDUCE_32(v_uint8x32, uchar, min, _mm_min_epu8)
OPENCV_HAL_IMPL_AVX_REDUCE_32(v_int8x32, schar, min, _mm_min_epi8)
OPENCV_HAL_IMPL_AVX_REDUCE_32(v_uint8x32, uchar, max, _mm_max_epu8)
OPENCV_HAL_IMPL_AVX_REDUCE_32(v_int8x32, schar, max, _mm_max_epi8)
#define OPENCV_HAL_IMPL_AVX_REDUCE_16(_Tpvec, sctype, func, intrin) \
inline sctype v_reduce_##func(const _Tpvec& a) \
{ \
@ -1062,31 +1090,6 @@ OPENCV_HAL_IMPL_AVX_REDUCE_8(v_int32x8, int, max, _mm_max_epi32)
OPENCV_HAL_IMPL_AVX_REDUCE_FLT(min, _mm_min_ps)
OPENCV_HAL_IMPL_AVX_REDUCE_FLT(max, _mm_max_ps)
inline ushort v_reduce_sum(const v_uint16x16& a)
{
__m128i a0 = _v256_extract_low(a.val);
__m128i a1 = _v256_extract_high(a.val);
__m128i s0 = _mm_adds_epu16(a0, a1);
s0 = _mm_adds_epu16(s0, _mm_srli_si128(s0, 8));
s0 = _mm_adds_epu16(s0, _mm_srli_si128(s0, 4));
s0 = _mm_adds_epu16(s0, _mm_srli_si128(s0, 2));
return (ushort)_mm_cvtsi128_si32(s0);
}
inline short v_reduce_sum(const v_int16x16& a)
{
__m256i s0 = _mm256_hadds_epi16(a.val, a.val);
s0 = _mm256_hadds_epi16(s0, s0);
s0 = _mm256_hadds_epi16(s0, s0);
__m128i s1 = _v256_extract_high(s0);
s1 = _mm_adds_epi16(_v256_extract_low(s0), s1);
return (short)_mm_cvtsi128_si32(s1);
}
inline int v_reduce_sum(const v_int32x8& a)
{
__m256i s0 = _mm256_hadd_epi32(a.val, a.val);
@ -1101,6 +1104,11 @@ inline int v_reduce_sum(const v_int32x8& a)
inline unsigned v_reduce_sum(const v_uint32x8& a)
{ return v_reduce_sum(v_reinterpret_as_s32(a)); }
inline int v_reduce_sum(const v_int16x16& a)
{ return v_reduce_sum(v_expand_low(a) + v_expand_high(a)); }
inline unsigned v_reduce_sum(const v_uint16x16& a)
{ return v_reduce_sum(v_expand_low(a) + v_expand_high(a)); }
inline float v_reduce_sum(const v_float32x8& a)
{
__m256 s0 = _mm256_hadd_ps(a.val, a.val);
@ -1112,6 +1120,18 @@ inline float v_reduce_sum(const v_float32x8& a)
return _mm_cvtss_f32(s1);
}
inline uint64 v_reduce_sum(const v_uint64x4& a)
{
uint64 CV_DECL_ALIGNED(32) idx[2];
_mm_store_si128((__m128i*)idx, _mm_add_epi64(_v256_extract_low(a.val), _v256_extract_high(a.val)));
return idx[0] + idx[1];
}
inline int64 v_reduce_sum(const v_int64x4& a)
{
int64 CV_DECL_ALIGNED(32) idx[2];
_mm_store_si128((__m128i*)idx, _mm_add_epi64(_v256_extract_low(a.val), _v256_extract_high(a.val)));
return idx[0] + idx[1];
}
inline double v_reduce_sum(const v_float64x4& a)
{
__m256d s0 = _mm256_hadd_pd(a.val, a.val);
@ -1166,26 +1186,39 @@ inline float v_reduce_sad(const v_float32x8& a, const v_float32x8& b)
}
/** Popcount **/
#define OPENCV_HAL_IMPL_AVX_POPCOUNT(_Tpvec) \
inline v_uint32x8 v_popcount(const _Tpvec& a) \
{ \
const v_uint32x8 m1 = v256_setall_u32(0x55555555); \
const v_uint32x8 m2 = v256_setall_u32(0x33333333); \
const v_uint32x8 m4 = v256_setall_u32(0x0f0f0f0f); \
v_uint32x8 p = v_reinterpret_as_u32(a); \
p = ((p >> 1) & m1) + (p & m1); \
p = ((p >> 2) & m2) + (p & m2); \
p = ((p >> 4) & m4) + (p & m4); \
p.val = _mm256_sad_epu8(p.val, _mm256_setzero_si256()); \
return p; \
}
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_uint8x32)
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_int8x32)
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_uint16x16)
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_int16x16)
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_uint32x8)
OPENCV_HAL_IMPL_AVX_POPCOUNT(v_int32x8)
inline v_uint8x32 v_popcount(const v_uint8x32& a)
{
__m256i _popcnt_table = _mm256_setr_epi8(0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4,
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4);
__m256i _popcnt_mask = _mm256_set1_epi8(0x0F);
return v_uint8x32(_mm256_add_epi8(_mm256_shuffle_epi8(_popcnt_table, _mm256_and_si256( a.val , _popcnt_mask)),
_mm256_shuffle_epi8(_popcnt_table, _mm256_and_si256(_mm256_srli_epi16(a.val, 4), _popcnt_mask))));
}
inline v_uint16x16 v_popcount(const v_uint16x16& a)
{
v_uint8x32 p = v_popcount(v_reinterpret_as_u8(a));
p += v_rotate_right<1>(p);
return v_reinterpret_as_u16(p) & v256_setall_u16(0x00ff);
}
inline v_uint32x8 v_popcount(const v_uint32x8& a)
{
v_uint8x32 p = v_popcount(v_reinterpret_as_u8(a));
p += v_rotate_right<1>(p);
p += v_rotate_right<2>(p);
return v_reinterpret_as_u32(p) & v256_setall_u32(0x000000ff);
}
inline v_uint64x4 v_popcount(const v_uint64x4& a)
{
return v_uint64x4(_mm256_sad_epu8(v_popcount(v_reinterpret_as_u8(a)).val, _mm256_setzero_si256()));
}
inline v_uint8x32 v_popcount(const v_int8x32& a)
{ return v_popcount(v_reinterpret_as_u8(a)); }
inline v_uint16x16 v_popcount(const v_int16x16& a)
{ return v_popcount(v_reinterpret_as_u16(a)); }
inline v_uint32x8 v_popcount(const v_int32x8& a)
{ return v_popcount(v_reinterpret_as_u32(a)); }
inline v_uint64x4 v_popcount(const v_int64x4& a)
{ return v_popcount(v_reinterpret_as_u64(a)); }
/** Mask **/
inline int v_signmask(const v_int8x32& a)

@ -603,27 +603,20 @@ static const unsigned char popCountTable[] =
3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8,
};
/** @brief Count the 1 bits in the vector and return 4 values
/** @brief Count the 1 bits in the vector lanes and return result as corresponding unsigned type
Scheme:
@code
{A1 A2 A3 ...} => popcount(A1)
{A1 A2 A3 ...} => {popcount(A1), popcount(A2), popcount(A3), ...}
@endcode
Any types but result will be in v_uint32x4*/
template<typename _Tp, int n> inline v_uint32x4 v_popcount(const v_reg<_Tp, n>& a)
For all integer types. */
template<typename _Tp, int n>
inline v_reg<typename V_TypeTraits<_Tp>::abs_type, n> v_popcount(const v_reg<_Tp, n>& a)
{
v_uint8x16 b;
b = v_reinterpret_as_u8(a);
for( int i = 0; i < v_uint8x16::nlanes; i++ )
{
b.s[i] = popCountTable[b.s[i]];
}
v_uint32x4 c;
for( int i = 0; i < v_uint32x4::nlanes; i++ )
{
c.s[i] = b.s[i*4] + b.s[i*4+1] + b.s[i*4+2] + b.s[i*4+3];
}
return c;
v_reg<typename V_TypeTraits<_Tp>::abs_type, n> b = v_reg<typename V_TypeTraits<_Tp>::abs_type, n>::zero();
for (int i = 0; i < (int)(n*sizeof(_Tp)); i++)
b.s[i/sizeof(_Tp)] += popCountTable[v_reinterpret_as_u8(a).s[i]];
return b;
}

@ -910,6 +910,31 @@ OPENCV_HAL_IMPL_NEON_LOADSTORE_OP(v_float32x4, float, f32)
OPENCV_HAL_IMPL_NEON_LOADSTORE_OP(v_float64x2, double, f64)
#endif
inline unsigned v_reduce_sum(const v_uint8x16& a)
{
uint32x4_t t0 = vpaddlq_u16(vpaddlq_u8(a.val));
uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
return vget_lane_u32(vpadd_u32(t1, t1), 0);
}
inline int v_reduce_sum(const v_int8x16& a)
{
int32x4_t t0 = vpaddlq_s16(vpaddlq_s8(a.val));
int32x2_t t1 = vpadd_s32(vget_low_s32(t0), vget_high_s32(t0));
return vget_lane_s32(vpadd_s32(t1, t1), 0);
}
inline unsigned v_reduce_sum(const v_uint16x8& a)
{
uint32x4_t t0 = vpaddlq_u16(a.val);
uint32x2_t t1 = vpadd_u32(vget_low_u32(t0), vget_high_u32(t0));
return vget_lane_u32(vpadd_u32(t1, t1), 0);
}
inline int v_reduce_sum(const v_int16x8& a)
{
int32x4_t t0 = vpaddlq_s16(a.val);
int32x2_t t1 = vpadd_s32(vget_low_s32(t0), vget_high_s32(t0));
return vget_lane_s32(vpadd_s32(t1, t1), 0);
}
#define OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(_Tpvec, _Tpnvec, scalartype, func, vectorfunc, suffix) \
inline scalartype v_reduce_##func(const _Tpvec& a) \
{ \
@ -918,12 +943,10 @@ inline scalartype v_reduce_##func(const _Tpvec& a) \
return (scalartype)vget_lane_##suffix(vp##vectorfunc##_##suffix(a0, a0),0); \
}
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_uint16x8, uint16x4, unsigned short, sum, add, u16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_uint16x8, uint16x4, unsigned short, max, max, u16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_uint16x8, uint16x4, unsigned short, min, min, u16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_int16x8, int16x4, short, sum, add, s16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_int16x8, int16x4, short, max, max, s16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_int16x8, int16x4, short, min, min, s16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_uint16x8, uint16x4, unsigned int, max, max, u16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_uint16x8, uint16x4, unsigned int, min, min, u16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_int16x8, int16x4, int, max, max, s16)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_8(v_int16x8, int16x4, int, min, min, s16)
#define OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(_Tpvec, _Tpnvec, scalartype, func, vectorfunc, suffix) \
inline scalartype v_reduce_##func(const _Tpvec& a) \
@ -942,6 +965,10 @@ OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, sum, add, f32)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, max, max, f32)
OPENCV_HAL_IMPL_NEON_REDUCE_OP_4(v_float32x4, float32x2, float, min, min, f32)
inline uint64 v_reduce_sum(const v_uint64x2& a)
{ return vget_lane_u64(vadd_u64(vget_low_u64(a.val), vget_high_u64(a.val)),0); }
inline int64 v_reduce_sum(const v_int64x2& a)
{ return vget_lane_s64(vadd_s64(vget_low_s64(a.val), vget_high_s64(a.val)),0); }
#if CV_SIMD128_64F
inline double v_reduce_sum(const v_float64x2& a)
{
@ -1007,21 +1034,22 @@ inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
return vget_lane_f32(vpadd_f32(t1, t1), 0);
}
#define OPENCV_HAL_IMPL_NEON_POPCOUNT(_Tpvec, cast) \
inline v_uint32x4 v_popcount(const _Tpvec& a) \
{ \
uint8x16_t t = vcntq_u8(cast(a.val)); \
uint16x8_t t0 = vpaddlq_u8(t); /* 16 -> 8 */ \
uint32x4_t t1 = vpaddlq_u16(t0); /* 8 -> 4 */ \
return v_uint32x4(t1); \
}
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_uint8x16, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_uint16x8, vreinterpretq_u8_u16)
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_uint32x4, vreinterpretq_u8_u32)
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_int8x16, vreinterpretq_u8_s8)
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_int16x8, vreinterpretq_u8_s16)
OPENCV_HAL_IMPL_NEON_POPCOUNT(v_int32x4, vreinterpretq_u8_s32)
inline v_uint8x16 v_popcount(const v_uint8x16& a)
{ return v_uint8x16(vcntq_u8(a.val)); }
inline v_uint8x16 v_popcount(const v_int8x16& a)
{ return v_uint8x16(vcntq_u8(vreinterpretq_u8_s8(a.val))); }
inline v_uint16x8 v_popcount(const v_uint16x8& a)
{ return v_uint16x8(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_u16(a.val)))); }
inline v_uint16x8 v_popcount(const v_int16x8& a)
{ return v_uint16x8(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_s16(a.val)))); }
inline v_uint32x4 v_popcount(const v_uint32x4& a)
{ return v_uint32x4(vpaddlq_u16(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_u32(a.val))))); }
inline v_uint32x4 v_popcount(const v_int32x4& a)
{ return v_uint32x4(vpaddlq_u16(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_s32(a.val))))); }
inline v_uint64x2 v_popcount(const v_uint64x2& a)
{ return v_uint64x2(vpaddlq_u32(vpaddlq_u16(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_u64(a.val)))))); }
inline v_uint64x2 v_popcount(const v_int64x2& a)
{ return v_uint64x2(vpaddlq_u32(vpaddlq_u16(vpaddlq_u8(vcntq_u8(vreinterpretq_u8_s64(a.val)))))); }
inline int v_signmask(const v_uint8x16& a)
{

@ -302,8 +302,8 @@ inline _Tpvec v_setall_##suffix(_Tp v) { return _Tpvec(_mm_set1_##ssuffix((_Tps)
template<typename _Tpvec0> inline _Tpvec v_reinterpret_as_##suffix(const _Tpvec0& a) \
{ return _Tpvec(cast(a.val)); }
OPENCV_HAL_IMPL_SSE_INITVEC(v_uint8x16, uchar, u8, si128, epi8, char, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_int8x16, schar, s8, si128, epi8, char, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_uint8x16, uchar, u8, si128, epi8, schar, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_int8x16, schar, s8, si128, epi8, schar, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_uint16x8, ushort, u16, si128, epi16, short, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_int16x8, short, s16, si128, epi16, short, OPENCV_HAL_NOP)
OPENCV_HAL_IMPL_SSE_INITVEC(v_uint32x4, unsigned, u32, si128, epi32, int, OPENCV_HAL_NOP)
@ -1393,6 +1393,41 @@ inline void v_store_high(_Tp* ptr, const _Tpvec& a) \
OPENCV_HAL_IMPL_SSE_LOADSTORE_FLT_OP(v_float32x4, float, ps)
OPENCV_HAL_IMPL_SSE_LOADSTORE_FLT_OP(v_float64x2, double, pd)
inline unsigned v_reduce_sum(const v_uint8x16& a)
{
__m128i half = _mm_sad_epu8(a.val, _mm_setzero_si128());
return (unsigned)_mm_cvtsi128_si32(_mm_add_epi32(half, _mm_unpackhi_epi64(half, half)));
}
inline int v_reduce_sum(const v_int8x16& a)
{
__m128i half = _mm_set1_epi8((schar)-128);
half = _mm_sad_epu8(_mm_xor_si128(a.val, half), _mm_setzero_si128());
return _mm_cvtsi128_si32(_mm_add_epi32(half, _mm_unpackhi_epi64(half, half))) - 2048;
}
#define OPENCV_HAL_IMPL_SSE_REDUCE_OP_16(func) \
inline schar v_reduce_##func(const v_int8x16& a) \
{ \
__m128i val = a.val; \
__m128i smask = _mm_set1_epi8((schar)-128); \
val = _mm_xor_si128(val, smask); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,8)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,4)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,2)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,1)); \
return (schar)_mm_cvtsi128_si32(val) ^ (schar)-128; \
} \
inline uchar v_reduce_##func(const v_uint8x16& a) \
{ \
__m128i val = a.val; \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,8)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,4)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,2)); \
val = _mm_##func##_epu8(val, _mm_srli_si128(val,1)); \
return (uchar)_mm_cvtsi128_si32(val); \
}
OPENCV_HAL_IMPL_SSE_REDUCE_OP_16(max)
OPENCV_HAL_IMPL_SSE_REDUCE_OP_16(min)
#define OPENCV_HAL_IMPL_SSE_REDUCE_OP_8(_Tpvec, scalartype, func, suffix, sbit) \
inline scalartype v_reduce_##func(const v_##_Tpvec& a) \
{ \
@ -1412,26 +1447,8 @@ inline unsigned scalartype v_reduce_##func(const v_u##_Tpvec& a) \
val = _mm_##func##_##suffix(val, _mm_srli_si128(val,2)); \
return (unsigned scalartype)(_mm_cvtsi128_si32(val) ^ sbit); \
}
#define OPENCV_HAL_IMPL_SSE_REDUCE_OP_8_SUM(_Tpvec, scalartype, suffix) \
inline scalartype v_reduce_sum(const v_##_Tpvec& a) \
{ \
__m128i val = a.val; \
val = _mm_adds_epi##suffix(val, _mm_srli_si128(val, 8)); \
val = _mm_adds_epi##suffix(val, _mm_srli_si128(val, 4)); \
val = _mm_adds_epi##suffix(val, _mm_srli_si128(val, 2)); \
return (scalartype)_mm_cvtsi128_si32(val); \
} \
inline unsigned scalartype v_reduce_sum(const v_u##_Tpvec& a) \
{ \
__m128i val = a.val; \
val = _mm_adds_epu##suffix(val, _mm_srli_si128(val, 8)); \
val = _mm_adds_epu##suffix(val, _mm_srli_si128(val, 4)); \
val = _mm_adds_epu##suffix(val, _mm_srli_si128(val, 2)); \
return (unsigned scalartype)_mm_cvtsi128_si32(val); \
}
OPENCV_HAL_IMPL_SSE_REDUCE_OP_8(int16x8, short, max, epi16, (short)-32768)
OPENCV_HAL_IMPL_SSE_REDUCE_OP_8(int16x8, short, min, epi16, (short)-32768)
OPENCV_HAL_IMPL_SSE_REDUCE_OP_8_SUM(int16x8, short, 16)
#define OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(_Tpvec, scalartype, regtype, suffix, cast_from, cast_to, extract) \
inline scalartype v_reduce_sum(const _Tpvec& a) \
@ -1456,6 +1473,23 @@ OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_uint32x4, unsigned, __m128i, epi32, OPENCV
OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_int32x4, int, __m128i, epi32, OPENCV_HAL_NOP, OPENCV_HAL_NOP, si128_si32)
OPENCV_HAL_IMPL_SSE_REDUCE_OP_4_SUM(v_float32x4, float, __m128, ps, _mm_castps_si128, _mm_castsi128_ps, ss_f32)
inline int v_reduce_sum(const v_int16x8& a)
{ return v_reduce_sum(v_expand_low(a) + v_expand_high(a)); }
inline unsigned v_reduce_sum(const v_uint16x8& a)
{ return v_reduce_sum(v_expand_low(a) + v_expand_high(a)); }
inline uint64 v_reduce_sum(const v_uint64x2& a)
{
uint64 CV_DECL_ALIGNED(32) idx[2];
v_store_aligned(idx, a);
return idx[0] + idx[1];
}
inline int64 v_reduce_sum(const v_int64x2& a)
{
int64 CV_DECL_ALIGNED(32) idx[2];
v_store_aligned(idx, a);
return idx[0] + idx[1];
}
inline double v_reduce_sum(const v_float64x2& a)
{
double CV_DECL_ALIGNED(32) idx[2];
@ -1520,27 +1554,42 @@ inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
return v_reduce_sum(v_absdiff(a, b));
}
#define OPENCV_HAL_IMPL_SSE_POPCOUNT(_Tpvec) \
inline v_uint32x4 v_popcount(const _Tpvec& a) \
{ \
__m128i m1 = _mm_set1_epi32(0x55555555); \
__m128i m2 = _mm_set1_epi32(0x33333333); \
__m128i m4 = _mm_set1_epi32(0x0f0f0f0f); \
__m128i p = a.val; \
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 1), m1), _mm_and_si128(p, m1)); \
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 2), m2), _mm_and_si128(p, m2)); \
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 4), m4), _mm_and_si128(p, m4)); \
p = _mm_adds_epi8(p, _mm_srli_si128(p, 1)); \
p = _mm_adds_epi8(p, _mm_srli_si128(p, 2)); \
return v_uint32x4(_mm_and_si128(p, _mm_set1_epi32(0x000000ff))); \
}
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_uint8x16)
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_uint16x8)
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_uint32x4)
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_int8x16)
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_int16x8)
OPENCV_HAL_IMPL_SSE_POPCOUNT(v_int32x4)
inline v_uint8x16 v_popcount(const v_uint8x16& a)
{
__m128i m1 = _mm_set1_epi32(0x55555555);
__m128i m2 = _mm_set1_epi32(0x33333333);
__m128i m4 = _mm_set1_epi32(0x0f0f0f0f);
__m128i p = a.val;
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 1), m1), _mm_and_si128(p, m1));
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 2), m2), _mm_and_si128(p, m2));
p = _mm_add_epi32(_mm_and_si128(_mm_srli_epi32(p, 4), m4), _mm_and_si128(p, m4));
return v_uint8x16(p);
}
inline v_uint16x8 v_popcount(const v_uint16x8& a)
{
v_uint8x16 p = v_popcount(v_reinterpret_as_u8(a));
p += v_rotate_right<1>(p);
return v_reinterpret_as_u16(p) & v_setall_u16(0x00ff);
}
inline v_uint32x4 v_popcount(const v_uint32x4& a)
{
v_uint8x16 p = v_popcount(v_reinterpret_as_u8(a));
p += v_rotate_right<1>(p);
p += v_rotate_right<2>(p);
return v_reinterpret_as_u32(p) & v_setall_u32(0x000000ff);
}
inline v_uint64x2 v_popcount(const v_uint64x2& a)
{
return v_uint64x2(_mm_sad_epu8(v_popcount(v_reinterpret_as_u8(a)).val, _mm_setzero_si128()));
}
inline v_uint8x16 v_popcount(const v_int8x16& a)
{ return v_popcount(v_reinterpret_as_u8(a)); }
inline v_uint16x8 v_popcount(const v_int16x8& a)
{ return v_popcount(v_reinterpret_as_u16(a)); }
inline v_uint32x4 v_popcount(const v_int32x4& a)
{ return v_popcount(v_reinterpret_as_u32(a)); }
inline v_uint64x2 v_popcount(const v_int64x2& a)
{ return v_popcount(v_reinterpret_as_u64(a)); }
#define OPENCV_HAL_IMPL_SSE_CHECK_SIGNS(_Tpvec, suffix, pack_op, and_op, signmask, allmask) \
inline int v_signmask(const _Tpvec& a) \

@ -692,15 +692,27 @@ inline _Tpvec v_extract(const _Tpvec& a, const _Tpvec& b)
////////// Reduce and mask /////////
/** Reduce **/
inline short v_reduce_sum(const v_int16x8& a)
inline uint v_reduce_sum(const v_uint8x16& a)
{
const vec_uint4 zero4 = vec_uint4_z;
vec_uint4 sum4 = vec_sum4s(a.val, zero4);
return (uint)vec_extract(vec_sums(vec_int4_c(sum4), vec_int4_c(zero4)), 3);
}
inline int v_reduce_sum(const v_int8x16& a)
{
const vec_int4 zero4 = vec_int4_z;
vec_int4 sum4 = vec_sum4s(a.val, zero4);
return (int)vec_extract(vec_sums(sum4, zero4), 3);
}
inline int v_reduce_sum(const v_int16x8& a)
{
const vec_int4 zero = vec_int4_z;
return saturate_cast<short>(vec_extract(vec_sums(vec_sum4s(a.val, zero), zero), 3));
return saturate_cast<int>(vec_extract(vec_sums(vec_sum4s(a.val, zero), zero), 3));
}
inline ushort v_reduce_sum(const v_uint16x8& a)
inline uint v_reduce_sum(const v_uint16x8& a)
{
const vec_int4 v4 = vec_int4_c(vec_unpackhu(vec_adds(a.val, vec_sld(a.val, a.val, 8))));
return saturate_cast<ushort>(vec_extract(vec_sums(v4, vec_int4_z), 3));
return saturate_cast<uint>(vec_extract(vec_sums(v4, vec_int4_z), 3));
}
#define OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(_Tpvec, _Tpvec2, scalartype, suffix, func) \
@ -719,6 +731,14 @@ OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, sum, vec_add)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, max, vec_max)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_4(v_float32x4, vec_float4, float, min, vec_min)
inline uint64 v_reduce_sum(const v_uint64x2& a)
{
return vec_extract(vec_add(a.val, vec_permi(a.val, a.val, 3)), 0);
}
inline int64 v_reduce_sum(const v_int64x2& a)
{
return vec_extract(vec_add(a.val, vec_permi(a.val, a.val, 3)), 0);
}
inline double v_reduce_sum(const v_float64x2& a)
{
return vec_extract(vec_add(a.val, vec_permi(a.val, a.val, 3)), 0);
@ -736,6 +756,19 @@ OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_uint16x8, vec_ushort8, ushort, min, vec_min)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_int16x8, vec_short8, short, max, vec_max)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_int16x8, vec_short8, short, min, vec_min)
#define OPENCV_HAL_IMPL_VSX_REDUCE_OP_16(_Tpvec, _Tpvec2, scalartype, suffix, func) \
inline scalartype v_reduce_##suffix(const _Tpvec& a) \
{ \
_Tpvec2 rs = func(a.val, vec_sld(a.val, a.val, 8)); \
rs = func(rs, vec_sld(rs, rs, 4)); \
rs = func(rs, vec_sld(rs, rs, 2)); \
return vec_extract(func(rs, vec_sld(rs, rs, 1)), 0); \
}
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_uint8x16, vec_uchar16, uchar, max, vec_max)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_uint8x16, vec_uchar16, uchar, min, vec_min)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_int8x16, vec_char16, schar, max, vec_max)
OPENCV_HAL_IMPL_VSX_REDUCE_OP_8(v_int8x16, vec_char16, schar, min, vec_min)
inline v_float32x4 v_reduce_sum4(const v_float32x4& a, const v_float32x4& b,
const v_float32x4& c, const v_float32x4& d)
{
@ -792,9 +825,22 @@ inline float v_reduce_sad(const v_float32x4& a, const v_float32x4& b)
}
/** Popcount **/
template<typename _Tpvec>
inline v_uint32x4 v_popcount(const _Tpvec& a)
{ return v_uint32x4(vec_popcntu(vec_uint4_c(a.val))); }
inline v_uint8x16 v_popcount(const v_uint8x16& a)
{ return v_uint8x16(vec_popcntu(a.val)); }
inline v_uint8x16 v_popcount(const v_int8x16& a)
{ return v_uint8x16(vec_popcntu(a.val)); }
inline v_uint16x8 v_popcount(const v_uint16x8& a)
{ return v_uint16x8(vec_popcntu(a.val)); }
inline v_uint16x8 v_popcount(const v_int16x8& a)
{ return v_uint16x8(vec_popcntu(a.val)); }
inline v_uint32x4 v_popcount(const v_uint32x4& a)
{ return v_uint32x4(vec_popcntu(a.val)); }
inline v_uint32x4 v_popcount(const v_int32x4& a)
{ return v_uint32x4(vec_popcntu(a.val)); }
inline v_uint64x2 v_popcount(const v_uint64x2& a)
{ return v_uint64x2(vec_popcntu(a.val)); }
inline v_uint64x2 v_popcount(const v_int64x2& a)
{ return v_uint64x2(vec_popcntu(a.val)); }
/** Mask **/
inline int v_signmask(const v_uint8x16& a)

@ -32,28 +32,15 @@ int normHamming(const uchar* a, int n)
int i = 0;
int result = 0;
#if CV_AVX2
{
__m256i _r0 = _mm256_setzero_si256();
__m256i _0 = _mm256_setzero_si256();
__m256i _popcnt_table = _mm256_setr_epi8(0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4,
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4);
__m256i _popcnt_mask = _mm256_set1_epi8(0x0F);
for(; i <= n - 32; i+= 32)
{
__m256i _a0 = _mm256_loadu_si256((const __m256i*)(a + i));
__m256i _popc0 = _mm256_shuffle_epi8(_popcnt_table, _mm256_and_si256(_a0, _popcnt_mask));
__m256i _popc1 = _mm256_shuffle_epi8(_popcnt_table,
_mm256_and_si256(_mm256_srli_epi16(_a0, 4), _popcnt_mask));
_r0 = _mm256_add_epi32(_r0, _mm256_sad_epu8(_0, _mm256_add_epi8(_popc0, _popc1)));
}
_r0 = _mm256_add_epi32(_r0, _mm256_shuffle_epi32(_r0, 2));
result = _mm256_extract_epi32_(_mm256_add_epi32(_r0, _mm256_permute2x128_si256(_r0, _r0, 1)), 0);
#if CV_SIMD && CV_SIMD_WIDTH > 16
{
v_uint64 t = vx_setzero_u64();
for (; i <= n - v_uint8::nlanes; i += v_uint8::nlanes)
t += v_popcount(v_reinterpret_as_u64(vx_load(a + i)));
result = (int)v_reduce_sum(t);
}
#endif // CV_AVX2
#endif
#if CV_POPCNT
{
@ -68,18 +55,14 @@ int normHamming(const uchar* a, int n)
result += CV_POPCNT_U32(*(uint*)(a + i));
}
}
#endif // CV_POPCNT
#if CV_SIMD128
#elif CV_SIMD
{
v_uint32x4 t = v_setzero_u32();
v_uint64x2 t = v_setzero_u64();
for(; i <= n - v_uint8x16::nlanes; i += v_uint8x16::nlanes)
{
t += v_popcount(v_load(a + i));
}
result += v_reduce_sum(t);
t += v_popcount(v_reinterpret_as_u64(v_load(a + i)));
result += (int)v_reduce_sum(t);
}
#endif // CV_SIMD128
#endif
#if CV_ENABLE_UNROLLED
for(; i <= n - 4; i += 4)
{
@ -100,31 +83,15 @@ int normHamming(const uchar* a, const uchar* b, int n)
int i = 0;
int result = 0;
#if CV_AVX2
{
__m256i _r0 = _mm256_setzero_si256();
__m256i _0 = _mm256_setzero_si256();
__m256i _popcnt_table = _mm256_setr_epi8(0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4,
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4);
__m256i _popcnt_mask = _mm256_set1_epi8(0x0F);
for(; i <= n - 32; i+= 32)
{
__m256i _a0 = _mm256_loadu_si256((const __m256i*)(a + i));
__m256i _b0 = _mm256_loadu_si256((const __m256i*)(b + i));
__m256i _xor = _mm256_xor_si256(_a0, _b0);
__m256i _popc0 = _mm256_shuffle_epi8(_popcnt_table, _mm256_and_si256(_xor, _popcnt_mask));
__m256i _popc1 = _mm256_shuffle_epi8(_popcnt_table,
_mm256_and_si256(_mm256_srli_epi16(_xor, 4), _popcnt_mask));
_r0 = _mm256_add_epi32(_r0, _mm256_sad_epu8(_0, _mm256_add_epi8(_popc0, _popc1)));
}
_r0 = _mm256_add_epi32(_r0, _mm256_shuffle_epi32(_r0, 2));
result = _mm256_extract_epi32_(_mm256_add_epi32(_r0, _mm256_permute2x128_si256(_r0, _r0, 1)), 0);
#if CV_SIMD && CV_SIMD_WIDTH > 16
{
v_uint64 t = vx_setzero_u64();
for (; i <= n - v_uint8::nlanes; i += v_uint8::nlanes)
t += v_popcount(v_reinterpret_as_u64(vx_load(a + i) ^ vx_load(b + i)));
result += (int)v_reduce_sum(t);
}
#endif // CV_AVX2
#endif
#if CV_POPCNT
{
@ -139,18 +106,14 @@ int normHamming(const uchar* a, const uchar* b, int n)
result += CV_POPCNT_U32(*(uint*)(a + i) ^ *(uint*)(b + i));
}
}
#endif // CV_POPCNT
#if CV_SIMD128
#elif CV_SIMD
{
v_uint32x4 t = v_setzero_u32();
v_uint64x2 t = v_setzero_u64();
for(; i <= n - v_uint8x16::nlanes; i += v_uint8x16::nlanes)
{
t += v_popcount(v_load(a + i) ^ v_load(b + i));
}
result += v_reduce_sum(t);
t += v_popcount(v_reinterpret_as_u64(vx_load(a + i) ^ vx_load(b + i)));
result += (int)v_reduce_sum(t);
}
#endif // CV_SIMD128
#endif
#if CV_ENABLE_UNROLLED
for(; i <= n - 4; i += 4)
{

@ -1195,7 +1195,7 @@ CV_IMPL const char* cvErrorStr( int status )
case CV_BadDepth : return "Input image depth is not supported by function";
case CV_StsUnmatchedFormats : return "Formats of input arguments do not match";
case CV_StsUnmatchedSizes : return "Sizes of input arguments do not match";
case CV_StsOutOfRange : return "One of arguments\' values is out of range";
case CV_StsOutOfRange : return "One of the arguments\' values is out of range";
case CV_StsUnsupportedFormat : return "Unsupported format or combination of formats";
case CV_BadCOI : return "Input COI is not supported";
case CV_BadNumChannels : return "Bad number of channels";

@ -686,18 +686,24 @@ template<typename R> struct TheTest
TheTest & test_popcount()
{
typedef typename V_RegTraits<R>::u_reg Ru;
static unsigned popcountTable[] = {
0, 1, 2, 4, 5, 7, 9, 12, 13, 15, 17, 20, 22, 25, 28, 32, 33,
35, 37, 40, 42, 45, 48, 52, 54, 57, 60, 64, 67, 71, 75, 80, 81,
83, 85, 88, 90, 93, 96, 100, 102, 105, 108, 112, 115, 119, 123,
128, 130, 133, 136, 140, 143, 147, 151, 156, 159, 163, 167, 172,
176, 181, 186, 192, 193
0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, //0x00-0x0f
1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, //0x10-0x1f
1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, //0x20-0x2f
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, //0x30-0x3f
1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, //0x40-0x4f
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, //0x50-0x5f
2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, //0x60-0x6f
3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, //0x70-0x7f
1 //0x80
};
Data<R> dataA;
R a = dataA;
unsigned resB = (unsigned)v_reduce_sum(v_popcount(a));
EXPECT_EQ(popcountTable[R::nlanes], resB);
Data<Ru> resB = v_popcount(a);
for (int i = 0; i < Ru::nlanes; ++i)
EXPECT_EQ(popcountTable[i + 1], resB[i]);
return *this;
}

@ -2794,9 +2794,6 @@ AsyncMat Net::forwardAsync(const String& outputName)
{
CV_TRACE_FUNCTION();
#ifdef CV_CXX11
if (impl->preferableBackend != DNN_BACKEND_INFERENCE_ENGINE)
CV_Error(Error::StsNotImplemented, "Asynchronous forward for backend which is different from DNN_BACKEND_INFERENCE_ENGINE");
String layerName = outputName;
if (layerName.empty())
@ -2805,6 +2802,9 @@ AsyncMat Net::forwardAsync(const String& outputName)
std::vector<LayerPin> pins(1, impl->getPinByAlias(layerName));
impl->setUpNet(pins);
if (impl->preferableBackend != DNN_BACKEND_INFERENCE_ENGINE)
CV_Error(Error::StsNotImplemented, "Asynchronous forward for backend which is different from DNN_BACKEND_INFERENCE_ENGINE");
impl->isAsync = true;
impl->forwardToLayer(impl->getLayerData(layerName));
impl->isAsync = false;

@ -312,15 +312,13 @@ public:
{
std::vector<UMat> inputs;
std::vector<UMat> outputs;
outs.getUMatVector(outputs);
bool use_half = (inps.depth() == CV_16S);
if (use_half)
{
std::vector<UMat> orig_inputs;
std::vector<UMat> orig_outputs;
inps.getUMatVector(orig_inputs);
outs.getUMatVector(orig_outputs);
inputs.resize(orig_inputs.size());
for (size_t i = 0; i < orig_inputs.size(); i++)
@ -329,7 +327,6 @@ public:
else
{
inps.getUMatVector(inputs);
outs.getUMatVector(outputs);
}
std::vector<LabelBBox> allDecodedBBoxes;
@ -362,19 +359,17 @@ public:
if (numKept == 0)
{
// Set confidences to zeros.
Range ranges[] = {Range::all(), Range::all(), Range::all(), Range(2, 3)};
if (use_half)
{
std::vector<UMat> orig_outputs;
outs.getUMatVector(orig_outputs);
orig_outputs[0](ranges).setTo(0);
} else
outputs[0](ranges).setTo(0);
outputs[0].setTo(0);
return true;
}
int outputShape[] = {1, 1, (int)numKept, 7};
UMat umat = UMat(4, outputShape, CV_32F);
UMat umat = use_half ? UMat::zeros(4, outputs[0].size, CV_32F) : outputs[0];
if (!use_half)
umat.setTo(0);
// If there are valid detections
if (numKept > 0)
{
Mat mat = umat.getMat(ACCESS_WRITE);
float* outputsData = mat.ptr<float>();
@ -393,16 +388,7 @@ public:
{
UMat half_umat;
convertFp16(umat, half_umat);
std::vector<UMat> orig_outputs;
outs.getUMatVector(orig_outputs);
orig_outputs.clear();
orig_outputs.push_back(half_umat);
outs.assign(orig_outputs);
} else {
outputs.clear();
outputs.push_back(umat);
outs.assign(outputs);
outs.assign(std::vector<UMat>(1, half_umat));
}
return true;
@ -484,15 +470,12 @@ public:
numKept += processDetections_(allDecodedBBoxes[i], allConfidenceScores[i], allIndices);
}
outputs[0].setTo(0);
// If there is no detections
if (numKept == 0)
{
// Set confidences to zeros.
Range ranges[] = {Range::all(), Range::all(), Range::all(), Range(2, 3)};
outputs[0](ranges).setTo(0);
return;
}
int outputShape[] = {1, 1, (int)numKept, 7};
outputs[0].create(4, outputShape, CV_32F);
float* outputsData = outputs[0].ptr<float>();
size_t count = 0;
@ -703,8 +686,6 @@ public:
prior_width += 1.0f;
prior_height += 1.0f;
}
CV_Assert(prior_width > 0);
CV_Assert(prior_height > 0);
float prior_center_x = prior_bbox.xmin + prior_width * .5;
float prior_center_y = prior_bbox.ymin + prior_height * .5;

@ -131,6 +131,9 @@ public:
CV_Assert(layerInternals.empty());
internals.push_back(layerOutputs[0]);
// Detections layer.
internals.push_back(shape(1, 1, keepTopAfterNMS, 7));
outputs.resize(2);
outputs[0] = shape(keepTopAfterNMS, 5);
outputs[1] = shape(keepTopAfterNMS, 1);
@ -176,13 +179,14 @@ public:
internals_.getUMatVector(internals);
CV_Assert(inputs.size() == 3);
CV_Assert(internals.size() == 3);
CV_Assert(internals.size() == 4);
const UMat& scores = inputs[0];
const UMat& bboxDeltas = inputs[1];
const UMat& imInfo = inputs[2];
UMat& priorBoxes = internals[0];
UMat& permuttedScores = internals[1];
UMat& permuttedDeltas = internals[2];
UMat& detections = internals[3];
CV_Assert(imInfo.total() >= 2);
// We've chosen the smallest data type because we need just a shape from it.
@ -217,7 +221,7 @@ public:
layerInputs[2] = priorBoxes;
layerInputs[3] = umat_fakeImageBlob;
layerOutputs[0] = UMat();
layerOutputs[0] = detections;
detectionOutputLayer->forward(layerInputs, layerOutputs, internals);
// DetectionOutputLayer produces 1x1xNx7 output where N might be less or
@ -237,10 +241,6 @@ public:
dst = outputs[1].rowRange(0, numDets);
layerOutputs[0].col(2).copyTo(dst);
if (numDets < keepTopAfterNMS)
for (int i = 0; i < 2; ++i)
outputs[i].rowRange(numDets, keepTopAfterNMS).setTo(0);
return true;
}
#endif
@ -266,13 +266,14 @@ public:
internals_arr.getMatVector(internals);
CV_Assert(inputs.size() == 3);
CV_Assert(internals.size() == 3);
CV_Assert(internals.size() == 4);
const Mat& scores = inputs[0];
const Mat& bboxDeltas = inputs[1];
const Mat& imInfo = inputs[2];
Mat& priorBoxes = internals[0];
Mat& permuttedScores = internals[1];
Mat& permuttedDeltas = internals[2];
Mat& detections = internals[3];
CV_Assert(imInfo.total() >= 2);
// We've chosen the smallest data type because we need just a shape from it.
@ -302,7 +303,7 @@ public:
layerInputs[2] = priorBoxes;
layerInputs[3] = fakeImageBlob;
layerOutputs[0] = Mat();
layerOutputs[0] = detections;
detectionOutputLayer->forward(layerInputs, layerOutputs, internals);
// DetectionOutputLayer produces 1x1xNx7 output where N might be less or
@ -319,10 +320,6 @@ public:
// The scores.
dst = outputs[1].rowRange(0, numDets);
layerOutputs[0].col(2).copyTo(dst);
if (numDets < keepTopAfterNMS)
for (int i = 0; i < 2; ++i)
outputs[i].rowRange(numDets, keepTopAfterNMS).setTo(0);
}
virtual Ptr<BackendNode> initInfEngine(const std::vector<Ptr<BackendWrapper> >&) CV_OVERRIDE

@ -172,7 +172,7 @@ TEST_P(DNNTestNetwork, MobileNet_SSD_Caffe)
Mat inp = blobFromImage(sample, 1.0f / 127.5, Size(300, 300), Scalar(127.5, 127.5, 127.5), false);
float diffScores = (target == DNN_TARGET_OPENCL_FP16 || target == DNN_TARGET_MYRIAD) ? 1.5e-2 : 0.0;
float diffSquares = (target == DNN_TARGET_MYRIAD) ? 0.063 : 0.0;
float detectionConfThresh = (target == DNN_TARGET_MYRIAD) ? 0.252 : 0.0;
float detectionConfThresh = (target == DNN_TARGET_MYRIAD) ? 0.252 : FLT_MIN;
processNet("dnn/MobileNetSSD_deploy.caffemodel", "dnn/MobileNetSSD_deploy.prototxt",
inp, "detection_out", "", diffScores, diffSquares, detectionConfThresh);
expectNoFallbacksFromIE(net);

@ -204,7 +204,7 @@ TEST(Reproducibility_SSD, Accuracy)
Mat out = net.forward("detection_out");
Mat ref = blobFromNPY(_tf("ssd_out.npy"));
normAssertDetections(ref, out);
normAssertDetections(ref, out, "", FLT_MIN);
}
typedef testing::TestWithParam<tuple<Backend, Target> > Reproducibility_MobileNet_SSD;
@ -225,6 +225,8 @@ TEST_P(Reproducibility_MobileNet_SSD, Accuracy)
net.setInput(inp);
Mat out = net.forward().clone();
ASSERT_EQ(out.size[2], 100);
const float scores_diff = (targetId == DNN_TARGET_OPENCL_FP16 || targetId == DNN_TARGET_MYRIAD) ? 1.5e-2 : 1e-5;
const float boxes_iou_diff = (targetId == DNN_TARGET_OPENCL_FP16 || targetId == DNN_TARGET_MYRIAD) ? 6.3e-2 : 1e-4;
Mat ref = blobFromNPY(_tf("mobilenet_ssd_caffe_out.npy"));

@ -341,11 +341,22 @@ public class JavaCamera2View extends CameraBridgeViewBase {
if (chromaPixelStride == 2) { // Chroma channels are interleaved
assert(planes[0].getPixelStride() == 1);
assert(planes[2].getPixelStride() == 2);
ByteBuffer y_plane = planes[0].getBuffer();
ByteBuffer uv_plane = planes[1].getBuffer();
ByteBuffer uv_plane1 = planes[1].getBuffer();
ByteBuffer uv_plane2 = planes[2].getBuffer();
Mat y_mat = new Mat(h, w, CvType.CV_8UC1, y_plane);
Mat uv_mat = new Mat(h / 2, w / 2, CvType.CV_8UC2, uv_plane);
Imgproc.cvtColorTwoPlane(y_mat, uv_mat, mRgba, Imgproc.COLOR_YUV2RGBA_NV21);
Mat uv_mat1 = new Mat(h / 2, w / 2, CvType.CV_8UC2, uv_plane1);
Mat uv_mat2 = new Mat(h / 2, w / 2, CvType.CV_8UC2, uv_plane2);
long addr_diff = uv_mat2.dataAddr() - uv_mat1.dataAddr();
if (addr_diff > 0) {
assert(addr_diff == 1);
Imgproc.cvtColorTwoPlane(y_mat, uv_mat1, mRgba, Imgproc.COLOR_YUV2RGBA_NV12);
} else {
assert(addr_diff == -1);
Imgproc.cvtColorTwoPlane(y_mat, uv_mat2, mRgba, Imgproc.COLOR_YUV2RGBA_NV21);
}
return mRgba;
} else { // Chroma channels are not interleaved
byte[] yuv_bytes = new byte[w*(h+h/2)];

@ -39,4 +39,4 @@ message(STATUS " include path: ${OpenCV_INCLUDE_DIRS}")
add_executable(${EXAMPLE_NAME} "${EXAMPLE_FILE}")
# Link your application with OpenCV libraries
target_link_libraries(${EXAMPLE_NAME} ${OpenCV_LIBS})
target_link_libraries(${EXAMPLE_NAME} LINK_PRIVATE ${OpenCV_LIBS})

@ -1,32 +1,3 @@
# Utility function: adds sample executable target with name "example_<group>_<file_name>"
# Usage:
# ocv_define_sample(<output target> <relative filename> <group>)
function(ocv_define_sample out_target source sub)
get_filename_component(name "${source}" NAME_WE)
set(the_target "example_${sub}_${name}")
add_executable(${the_target} "${source}")
set_target_properties(${the_target} PROPERTIES PROJECT_LABEL "(sample) ${name}")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples/${sub}")
endif()
if(WIN32 AND MSVC AND NOT BUILD_SHARED_LIBS)
set_target_properties(${the_target} PROPERTIES LINK_FLAGS "/NODEFAULTLIB:atlthunk.lib /NODEFAULTLIB:atlsd.lib /DEBUG")
endif()
if(WIN32)
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/${sub}" COMPONENT samples)
endif()
# Add single target to build all samples in the group: 'make opencv_samples_cpp'
set(parent_target opencv_samples_${sub})
if(NOT TARGET ${parent_target})
add_custom_target(${parent_target})
if(TARGET opencv_samples)
add_dependencies(opencv_samples ${parent_target})
endif()
endif()
add_dependencies(${parent_target} ${the_target})
set(${out_target} ${the_target} PARENT_SCOPE)
endfunction()
if(NOT CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_LIST_DIR)
#===================================================================================================
#
@ -34,6 +5,8 @@ if(NOT CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_LIST_DIR)
#
#===================================================================================================
include("${CMAKE_CURRENT_LIST_DIR}/samples_utils.cmake")
function(ocv_install_example_src relpath)
if(INSTALL_C_EXAMPLES)
file(GLOB files ${ARGN})
@ -43,6 +16,10 @@ function(ocv_install_example_src relpath)
endif()
endfunction()
if((TARGET Threads::Threads OR HAVE_PTHREAD OR MSVC OR APPLE) AND NOT OPENCV_EXAMPLES_DISABLE_THREADS)
add_definitions(-DHAVE_THREADS=1)
endif()
add_subdirectory(cpp)
add_subdirectory(java/tutorial_code)
add_subdirectory(dnn)
@ -98,6 +75,8 @@ option(BUILD_EXAMPLES "Build samples" ON)
#    cpp/
find_package(OpenCV REQUIRED PATHS "..")
include("${CMAKE_CURRENT_LIST_DIR}/samples_utils.cmake")
function(ocv_install_example_src)
# not used in this branch
endfunction()
@ -129,6 +108,17 @@ endif()
add_definitions(-DDISABLE_OPENCV_24_COMPATIBILITY=1) # Avoid C-like legacy API
if(OPENCV_EXAMPLES_DISABLE_THREADS)
# nothing
elseif(MSVC OR APPLE)
set(HAVE_THREADS 1)
else()
find_package(Threads)
endif()
if((TARGET Threads::Threads OR HAVE_THREADS) AND NOT OPENCV_EXAMPLES_DISABLE_THREADS)
add_definitions(-DHAVE_THREADS=1)
endif()
add_subdirectory(cpp)
if(WIN32)
add_subdirectory(directx)

@ -35,12 +35,12 @@ foreach(sample_filename ${cpp_samples})
set(package "tutorial")
endif()
ocv_define_sample(tgt ${sample_filename} ${package})
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
if(sample_filename MATCHES "/gpu/" AND HAVE_opencv_cudaarithm AND HAVE_opencv_cuda_filters)
ocv_target_link_libraries(${tgt} opencv_cudaarithm opencv_cudafilters)
ocv_target_link_libraries(${tgt} LINK_PRIVATE opencv_cudaarithm opencv_cudafilters)
endif()
if(sample_filename MATCHES "/viz/")
ocv_target_link_libraries(${tgt} ${VTK_LIBRARIES})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${VTK_LIBRARIES})
target_compile_definitions(${tgt} PRIVATE -DUSE_VTK)
endif()
if(HAVE_OPENGL AND sample_filename MATCHES "detect_mser")

@ -27,4 +27,4 @@ message(STATUS " include path: ${OpenCV_INCLUDE_DIRS}")
add_executable(opencv_example example.cpp)
# Link your application with OpenCV libraries
target_link_libraries(opencv_example ${OpenCV_LIBS})
target_link_libraries(opencv_example LINK_PRIVATE ${OpenCV_LIBS})

@ -17,5 +17,5 @@ ocv_include_modules_recurse(${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
add_executable( ${target}pnp_registration ${sample_dir}main_registration.cpp ${sample_pnplib} )
add_executable( ${target}pnp_detection ${sample_dir}main_detection.cpp ${sample_pnplib} )
ocv_target_link_libraries( ${target}pnp_registration ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS} )
ocv_target_link_libraries( ${target}pnp_detection ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS} )
ocv_target_link_libraries(${target}pnp_registration LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${target}pnp_detection LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_CPP_SAMPLES_REQUIRED_DEPS})

@ -0,0 +1,86 @@
#include <iostream>
#include <opencv2/imgcodecs.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/video.hpp>
using namespace cv;
using namespace std;
int main(int argc, char **argv)
{
const string about =
"This sample demonstrates the camshift algorithm.\n"
"The example file can be downloaded from:\n"
" https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4";
const string keys =
"{ h help | | print this help message }"
"{ @image |<none>| path to image file }";
CommandLineParser parser(argc, argv, keys);
parser.about(about);
if (parser.has("help"))
{
parser.printMessage();
return 0;
}
string filename = parser.get<string>("@image");
if (!parser.check())
{
parser.printErrors();
return 0;
}
VideoCapture capture(filename);
if (!capture.isOpened()){
//error in opening the video input
cerr << "Unable to open file!" << endl;
return 0;
}
Mat frame, roi, hsv_roi, mask;
// take first frame of the video
capture >> frame;
// setup initial location of window
Rect track_window(300, 200, 100, 50); // simply hardcoded the values
// set up the ROI for tracking
roi = frame(track_window);
cvtColor(roi, hsv_roi, COLOR_BGR2HSV);
inRange(hsv_roi, Scalar(0, 60, 32), Scalar(180, 255, 255), mask);
float range_[] = {0, 180};
const float* range[] = {range_};
Mat roi_hist;
int histSize[] = {180};
int channels[] = {0};
calcHist(&hsv_roi, 1, channels, mask, roi_hist, 1, histSize, range);
normalize(roi_hist, roi_hist, 0, 255, NORM_MINMAX);
// Setup the termination criteria, either 10 iteration or move by atleast 1 pt
TermCriteria term_crit(TermCriteria::EPS | TermCriteria::COUNT, 10, 1);
while(true){
Mat hsv, dst;
capture >> frame;
if (frame.empty())
break;
cvtColor(frame, hsv, COLOR_BGR2HSV);
calcBackProject(&hsv, 1, channels, roi_hist, dst, range);
// apply camshift to get the new location
RotatedRect rot_rect = CamShift(dst, track_window, term_crit);
// Draw it on image
Point2f points[4];
rot_rect.points(points);
for (int i = 0; i < 4; i++)
line(frame, points[i], points[(i+1)%4], 255, 2);
imshow("img2", frame);
int keyboard = waitKey(30);
if (keyboard == 'q' || keyboard == 27)
break;
}
}

@ -0,0 +1,83 @@
#include <iostream>
#include <opencv2/imgcodecs.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/video.hpp>
using namespace cv;
using namespace std;
int main(int argc, char **argv)
{
const string about =
"This sample demonstrates the meanshift algorithm.\n"
"The example file can be downloaded from:\n"
" https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4";
const string keys =
"{ h help | | print this help message }"
"{ @image |<none>| path to image file }";
CommandLineParser parser(argc, argv, keys);
parser.about(about);
if (parser.has("help"))
{
parser.printMessage();
return 0;
}
string filename = parser.get<string>("@image");
if (!parser.check())
{
parser.printErrors();
return 0;
}
VideoCapture capture(filename);
if (!capture.isOpened()){
//error in opening the video input
cerr << "Unable to open file!" << endl;
return 0;
}
Mat frame, roi, hsv_roi, mask;
// take first frame of the video
capture >> frame;
// setup initial location of window
Rect track_window(300, 200, 100, 50); // simply hardcoded the values
// set up the ROI for tracking
roi = frame(track_window);
cvtColor(roi, hsv_roi, COLOR_BGR2HSV);
inRange(hsv_roi, Scalar(0, 60, 32), Scalar(180, 255, 255), mask);
float range_[] = {0, 180};
const float* range[] = {range_};
Mat roi_hist;
int histSize[] = {180};
int channels[] = {0};
calcHist(&hsv_roi, 1, channels, mask, roi_hist, 1, histSize, range);
normalize(roi_hist, roi_hist, 0, 255, NORM_MINMAX);
// Setup the termination criteria, either 10 iteration or move by atleast 1 pt
TermCriteria term_crit(TermCriteria::EPS | TermCriteria::COUNT, 10, 1);
while(true){
Mat hsv, dst;
capture >> frame;
if (frame.empty())
break;
cvtColor(frame, hsv, COLOR_BGR2HSV);
calcBackProject(&hsv, 1, channels, roi_hist, dst, range);
// apply meanshift to get the new location
meanShift(dst, track_window, term_crit);
// Draw it on image
rectangle(frame, track_window, 255, 2);
imshow("img2", frame);
int keyboard = waitKey(30);
if (keyboard == 'q' || keyboard == 27)
break;
}
}

@ -0,0 +1,101 @@
#include <iostream>
#include <opencv2/core.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/video.hpp>
using namespace cv;
using namespace std;
int main(int argc, char **argv)
{
const string about =
"This sample demonstrates Lucas-Kanade Optical Flow calculation.\n"
"The example file can be downloaded from:\n"
" https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4";
const string keys =
"{ h help | | print this help message }"
"{ @image |<none>| path to image file }";
CommandLineParser parser(argc, argv, keys);
parser.about(about);
if (parser.has("help"))
{
parser.printMessage();
return 0;
}
string filename = parser.get<string>("@image");
if (!parser.check())
{
parser.printErrors();
return 0;
}
VideoCapture capture(filename);
if (!capture.isOpened()){
//error in opening the video input
cerr << "Unable to open file!" << endl;
return 0;
}
// Create some random colors
vector<Scalar> colors;
RNG rng;
for(int i = 0; i < 100; i++)
{
int r = rng.uniform(0, 256);
int g = rng.uniform(0, 256);
int b = rng.uniform(0, 256);
colors.push_back(Scalar(r,g,b));
}
Mat old_frame, old_gray;
vector<Point2f> p0, p1;
// Take first frame and find corners in it
capture >> old_frame;
cvtColor(old_frame, old_gray, COLOR_BGR2GRAY);
goodFeaturesToTrack(old_gray, p0, 100, 0.3, 7, Mat(), 7, false, 0.04);
// Create a mask image for drawing purposes
Mat mask = Mat::zeros(old_frame.size(), old_frame.type());
while(true){
Mat frame, frame_gray;
capture >> frame;
if (frame.empty())
break;
cvtColor(frame, frame_gray, COLOR_BGR2GRAY);
// calculate optical flow
vector<uchar> status;
vector<float> err;
TermCriteria criteria = TermCriteria((TermCriteria::COUNT) + (TermCriteria::EPS), 10, 0.03);
calcOpticalFlowPyrLK(old_gray, frame_gray, p0, p1, status, err, Size(15,15), 2, criteria);
vector<Point2f> good_new;
for(uint i = 0; i < p0.size(); i++)
{
// Select good points
if(status[i] == 1) {
good_new.push_back(p1[i]);
// draw the tracks
line(mask,p1[i], p0[i], colors[i], 2);
circle(frame, p1[i], 5, colors[i], -1);
}
}
Mat img;
add(frame, mask, img);
imshow("Frame", img);
int keyboard = waitKey(30);
if (keyboard == 'q' || keyboard == 27)
break;
// Now update the previous frame and previous points
old_gray = frame_gray.clone();
p0 = good_new;
}
}

@ -0,0 +1,59 @@
#include <iostream>
#include <opencv2/core.hpp>
#include <opencv2/highgui.hpp>
#include <opencv2/imgproc.hpp>
#include <opencv2/videoio.hpp>
#include <opencv2/video.hpp>
using namespace cv;
using namespace std;
int main()
{
VideoCapture capture(samples::findFile("vtest.avi"));
if (!capture.isOpened()){
//error in opening the video input
cerr << "Unable to open file!" << endl;
return 0;
}
Mat frame1, prvs;
capture >> frame1;
cvtColor(frame1, prvs, COLOR_BGR2GRAY);
while(true){
Mat frame2, next;
capture >> frame2;
if (frame2.empty())
break;
cvtColor(frame2, next, COLOR_BGR2GRAY);
Mat flow(prvs.size(), CV_32FC2);
calcOpticalFlowFarneback(prvs, next, flow, 0.5, 3, 15, 3, 5, 1.2, 0);
// visualization
Mat flow_parts[2];
split(flow, flow_parts);
Mat magnitude, angle, magn_norm;
cartToPolar(flow_parts[0], flow_parts[1], magnitude, angle, true);
normalize(magnitude, magn_norm, 0.0f, 1.0f, NORM_MINMAX);
angle *= ((1.f / 360.f) * (180.f / 255.f));
//build hsv image
Mat _hsv[3], hsv, hsv8, bgr;
_hsv[0] = angle;
_hsv[1] = Mat::ones(angle.size(), CV_32F);
_hsv[2] = magn_norm;
merge(_hsv, 3, hsv);
hsv.convertTo(hsv8, CV_8U, 255.0);
cvtColor(hsv8, bgr, COLOR_HSV2BGR);
imshow("frame2", bgr);
int keyboard = waitKey(30);
if (keyboard == 'q' || keyboard == 27)
break;
prvs = next;
}
}

@ -17,5 +17,5 @@ ocv_include_modules_recurse(${tgt} ${OPENCV_DIRECTX_SAMPLES_REQUIRED_DEPS})
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} directx)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_DIRECTX_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_DIRECTX_SAMPLES_REQUIRED_DEPS})
endforeach()

@ -18,5 +18,5 @@ ocv_include_modules_recurse(${OPENCV_DNN_SAMPLES_REQUIRED_DEPS})
file(GLOB_RECURSE dnn_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${dnn_samples})
ocv_define_sample(tgt ${sample_filename} dnn)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_DNN_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_DNN_SAMPLES_REQUIRED_DEPS})
endforeach()

@ -5,6 +5,11 @@
#include <opencv2/imgproc.hpp>
#include <opencv2/highgui.hpp>
#ifdef CV_CXX11
#include <thread>
#include <queue>
#endif
#include "common.hpp"
std::string keys =
@ -26,8 +31,9 @@ std::string keys =
"0: CPU target (by default), "
"1: OpenCL, "
"2: OpenCL fp16 (half-float precision), "
"3: VPU }";
"3: VPU }"
"{ async | 0 | Number of asynchronous forwards at the same time. "
"Choose 0 for synchronous mode }";
using namespace cv;
using namespace dnn;
@ -35,13 +41,66 @@ using namespace dnn;
float confThreshold, nmsThreshold;
std::vector<std::string> classes;
inline void preprocess(const Mat& frame, Net& net, Size inpSize, float scale,
const Scalar& mean, bool swapRB);
void postprocess(Mat& frame, const std::vector<Mat>& out, Net& net);
void drawPred(int classId, float conf, int left, int top, int right, int bottom, Mat& frame);
void callback(int pos, void* userdata);
std::vector<String> getOutputsNames(const Net& net);
#ifdef CV_CXX11
template <typename T>
class QueueFPS : public std::queue<T>
{
public:
QueueFPS() : counter(0) {}
void push(const T& entry)
{
std::lock_guard<std::mutex> lock(mutex);
std::queue<T>::push(entry);
counter += 1;
if (counter == 1)
{
// Start counting from a second frame (warmup).
tm.reset();
tm.start();
}
}
T get()
{
std::lock_guard<std::mutex> lock(mutex);
T entry = this->front();
this->pop();
return entry;
}
float getFPS()
{
tm.stop();
double fps = counter / tm.getTimeSec();
tm.start();
return static_cast<float>(fps);
}
void clear()
{
std::lock_guard<std::mutex> lock(mutex);
while (!this->empty())
this->pop();
}
unsigned int counter;
private:
TickMeter tm;
std::mutex mutex;
};
#endif // CV_CXX11
int main(int argc, char** argv)
{
@ -67,6 +126,7 @@ int main(int argc, char** argv)
bool swapRB = parser.get<bool>("rgb");
int inpWidth = parser.get<int>("width");
int inpHeight = parser.get<int>("height");
size_t async = parser.get<int>("async");
CV_Assert(parser.has("model"));
std::string modelPath = findFile(parser.get<String>("model"));
std::string configPath = findFile(parser.get<String>("config"));
@ -104,6 +164,108 @@ int main(int argc, char** argv)
else
cap.open(parser.get<int>("device"));
#ifdef CV_CXX11
bool process = true;
// Frames capturing thread
QueueFPS<Mat> framesQueue;
std::thread framesThread([&](){
Mat frame;
while (process)
{
cap >> frame;
if (!frame.empty())
framesQueue.push(frame.clone());
else
break;
}
});
// Frames processing thread
QueueFPS<Mat> processedFramesQueue;
QueueFPS<std::vector<Mat> > predictionsQueue;
std::thread processingThread([&](){
std::queue<std::future<Mat> > futureOutputs;
Mat blob;
while (process)
{
// Get a next frame
Mat frame;
{
if (!framesQueue.empty())
{
frame = framesQueue.get();
if (async)
{
if (futureOutputs.size() == async)
frame = Mat();
}
else
framesQueue.clear(); // Skip the rest of frames
}
}
// Process the frame
if (!frame.empty())
{
preprocess(frame, net, Size(inpWidth, inpHeight), scale, mean, swapRB);
processedFramesQueue.push(frame);
if (async)
{
futureOutputs.push(net.forwardAsync());
}
else
{
std::vector<Mat> outs;
net.forward(outs, outNames);
predictionsQueue.push(outs);
}
}
while (!futureOutputs.empty() &&
futureOutputs.front().wait_for(std::chrono::seconds(0)) == std::future_status::ready)
{
Mat out = futureOutputs.front().get();
predictionsQueue.push({out});
futureOutputs.pop();
}
}
});
// Postprocessing and rendering loop
while (waitKey(1) < 0)
{
if (predictionsQueue.empty())
continue;
std::vector<Mat> outs = predictionsQueue.get();
Mat frame = processedFramesQueue.get();
postprocess(frame, outs, net);
if (predictionsQueue.counter > 1)
{
std::string label = format("Camera: %.2f FPS", framesQueue.getFPS());
putText(frame, label, Point(0, 15), FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 255, 0));
label = format("Network: %.2f FPS", predictionsQueue.getFPS());
putText(frame, label, Point(0, 30), FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 255, 0));
label = format("Skipped frames: %d", framesQueue.counter - predictionsQueue.counter);
putText(frame, label, Point(0, 45), FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 255, 0));
}
imshow(kWinName, frame);
}
process = false;
framesThread.join();
processingThread.join();
#else // CV_CXX11
if (async)
CV_Error(Error::StsNotImplemented, "Asynchronous forward is supported only with Inference Engine backend.");
// Process frames.
Mat frame, blob;
while (waitKey(1) < 0)
@ -115,19 +277,8 @@ int main(int argc, char** argv)
break;
}
// Create a 4D blob from a frame.
Size inpSize(inpWidth > 0 ? inpWidth : frame.cols,
inpHeight > 0 ? inpHeight : frame.rows);
blobFromImage(frame, blob, scale, inpSize, mean, swapRB, false);
preprocess(frame, net, Size(inpWidth, inpHeight), scale, mean, swapRB);
// Run a model.
net.setInput(blob);
if (net.getLayer(0)->outputNameToIndex("im_info") != -1) // Faster-RCNN or R-FCN
{
resize(frame, frame, inpSize);
Mat imInfo = (Mat_<float>(1, 3) << inpSize.height, inpSize.width, 1.6f);
net.setInput(imInfo, "im_info");
}
std::vector<Mat> outs;
net.forward(outs, outNames);
@ -142,9 +293,29 @@ int main(int argc, char** argv)
imshow(kWinName, frame);
}
#endif // CV_CXX11
return 0;
}
inline void preprocess(const Mat& frame, Net& net, Size inpSize, float scale,
const Scalar& mean, bool swapRB)
{
static Mat blob;
// Create a 4D blob from a frame.
if (inpSize.width <= 0) inpSize.width = frame.cols;
if (inpSize.height <= 0) inpSize.height = frame.rows;
blobFromImage(frame, blob, 1.0, inpSize, Scalar(), swapRB, false, CV_8U);
// Run a model.
net.setInput(blob, "", scale, mean);
if (net.getLayer(0)->outputNameToIndex("im_info") != -1) // Faster-RCNN or R-FCN
{
resize(frame, frame, inpSize);
Mat imInfo = (Mat_<float>(1, 3) << inpSize.height, inpSize.width, 1.6f);
net.setInput(imInfo, "im_info");
}
}
void postprocess(Mat& frame, const std::vector<Mat>& outs, Net& net)
{
static std::vector<int> outLayers = net.getUnconnectedOutLayers();

@ -1,6 +1,13 @@
import cv2 as cv
import argparse
import numpy as np
import sys
import time
from threading import Thread
if sys.version_info[0] == '2':
import Queue as queue
else:
import queue
from common import *
from tf_text_graph_common import readTextMessage
@ -35,6 +42,9 @@ parser.add_argument('--target', choices=targets, default=cv.dnn.DNN_TARGET_CPU,
'%d: OpenCL, '
'%d: OpenCL fp16 (half-float precision), '
'%d: VPU' % targets)
parser.add_argument('--async', type=int, default=0,
help='Number of asynchronous forwards at the same time. '
'Choose 0 for synchronous mode')
args, _ = parser.parse_known_args()
add_preproc_args(args.zoo, parser, 'object_detection')
parser = argparse.ArgumentParser(parents=[parser],
@ -173,32 +183,125 @@ def callback(pos):
cv.createTrackbar('Confidence threshold, %', winName, int(confThreshold * 100), 99, callback)
cap = cv.VideoCapture(cv.samples.findFileOrKeep(args.input) if args.input else 0)
class QueueFPS(queue.Queue):
def __init__(self):
queue.Queue.__init__(self)
self.startTime = 0
self.counter = 0
def put(self, v):
queue.Queue.put(self, v)
self.counter += 1
if self.counter == 1:
self.startTime = time.time()
def getFPS(self):
return self.counter / (time.time() - self.startTime)
process = True
#
# Frames capturing thread
#
framesQueue = QueueFPS()
def framesThreadBody():
global framesQueue, process
while process:
hasFrame, frame = cap.read()
if not hasFrame:
break
framesQueue.put(frame)
#
# Frames processing thread
#
processedFramesQueue = queue.Queue()
predictionsQueue = QueueFPS()
def processingThreadBody():
global processedFramesQueue, predictionsQueue, args, process
futureOutputs = []
while process:
# Get a next frame
frame = None
try:
frame = framesQueue.get_nowait()
if args.async:
if len(futureOutputs) == args.async:
frame = None # Skip the frame
else:
framesQueue.queue.clear() # Skip the rest of frames
except queue.Empty:
pass
if not frame is None:
frameHeight = frame.shape[0]
frameWidth = frame.shape[1]
# Create a 4D blob from a frame.
inpWidth = args.width if args.width else frameWidth
inpHeight = args.height if args.height else frameHeight
blob = cv.dnn.blobFromImage(frame, size=(inpWidth, inpHeight), swapRB=args.rgb, ddepth=cv.CV_8U)
processedFramesQueue.put(frame)
# Run a model
net.setInput(blob, scalefactor=args.scale, mean=args.mean)
if net.getLayer(0).outputNameToIndex('im_info') != -1: # Faster-RCNN or R-FCN
frame = cv.resize(frame, (inpWidth, inpHeight))
net.setInput(np.array([[inpHeight, inpWidth, 1.6]], dtype=np.float32), 'im_info')
if args.async:
futureOutputs.append(net.forwardAsync())
else:
outs = net.forward(outNames)
predictionsQueue.put(np.copy(outs))
while futureOutputs and futureOutputs[0].wait_for(0) == 0:
out = futureOutputs[0].get()
predictionsQueue.put(np.copy([out]))
del futureOutputs[0]
framesThread = Thread(target=framesThreadBody)
framesThread.start()
processingThread = Thread(target=processingThreadBody)
processingThread.start()
#
# Postprocessing and rendering loop
#
while cv.waitKey(1) < 0:
hasFrame, frame = cap.read()
if not hasFrame:
cv.waitKey()
break
try:
# Request prediction first because they put after frames
outs = predictionsQueue.get_nowait()
frame = processedFramesQueue.get_nowait()
frameHeight = frame.shape[0]
frameWidth = frame.shape[1]
postprocess(frame, outs)
# Put efficiency information.
if predictionsQueue.counter > 1:
label = 'Camera: %.2f FPS' % (framesQueue.getFPS())
cv.putText(frame, label, (0, 15), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0))
# Create a 4D blob from a frame.
inpWidth = args.width if args.width else frameWidth
inpHeight = args.height if args.height else frameHeight
blob = cv.dnn.blobFromImage(frame, args.scale, (inpWidth, inpHeight), args.mean, args.rgb, crop=False)
label = 'Network: %.2f FPS' % (predictionsQueue.getFPS())
cv.putText(frame, label, (0, 30), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0))
# Run a model
net.setInput(blob)
if net.getLayer(0).outputNameToIndex('im_info') != -1: # Faster-RCNN or R-FCN
frame = cv.resize(frame, (inpWidth, inpHeight))
net.setInput(np.array([[inpHeight, inpWidth, 1.6]], dtype=np.float32), 'im_info')
outs = net.forward(outNames)
label = 'Skipped frames: %d' % (framesQueue.counter - predictionsQueue.counter)
cv.putText(frame, label, (0, 45), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0))
postprocess(frame, outs)
cv.imshow(winName, frame)
except queue.Empty:
pass
# Put efficiency information.
t, _ = net.getPerfProfile()
label = 'Inference time: %.2f ms' % (t * 1000.0 / cv.getTickFrequency())
cv.putText(frame, label, (0, 15), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0))
cv.imshow(winName, frame)
process = False
framesThread.join()
processingThread.join()

@ -51,11 +51,11 @@ endif()
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} gpu)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_CUDA_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_CUDA_SAMPLES_REQUIRED_DEPS})
if(HAVE_opencv_xfeatures2d)
ocv_target_link_libraries(${tgt} opencv_xfeatures2d)
ocv_target_link_libraries(${tgt} LINK_PRIVATE opencv_xfeatures2d)
endif()
if(HAVE_opencv_cudacodec)
ocv_target_link_libraries(${tgt} opencv_cudacodec)
ocv_target_link_libraries(${tgt} LINK_PRIVATE opencv_cudacodec)
endif()
endforeach()

@ -31,7 +31,7 @@ ocv_include_directories(${OpenCL_INCLUDE_DIR})
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} opencl)
ocv_target_link_libraries(${tgt}
ocv_target_link_libraries(${tgt} LINK_PRIVATE
${OPENCV_LINKER_LIBS}
${OPENCV_OPENCL_SAMPLES_REQUIRED_DEPS}
${OpenCL_LIBRARY})

@ -23,9 +23,9 @@ if(BUILD_EXAMPLES AND OCV_DEPENDENCIES_FOUND)
endif()
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} opengl)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_OPENGL_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_OPENGL_SAMPLES_REQUIRED_DEPS})
if(sample_filename STREQUAL "opengl_interop.cpp")
ocv_target_link_libraries(${tgt} ${X11_LIBRARIES})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${X11_LIBRARIES})
ocv_target_include_directories(${tgt} ${X11_INCLUDE_DIR})
endif()
endforeach()

@ -21,5 +21,5 @@ add_definitions(-DIVX_HIDE_INFO_WARNINGS)
file(GLOB_RECURSE cpp_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${cpp_samples})
ocv_define_sample(tgt ${sample_filename} openvx)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_OPENVX_SAMPLE_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_OPENVX_SAMPLE_REQUIRED_DEPS})
endforeach()

@ -0,0 +1,50 @@
import numpy as np
import cv2 as cv
import argparse
parser = argparse.ArgumentParser(description='This sample demonstrates the camshift algorithm. \
The example file can be downloaded from: \
https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4')
parser.add_argument('image', type=str, help='path to image file')
args = parser.parse_args()
cap = cv.VideoCapture(args.image)
# take first frame of the video
ret,frame = cap.read()
# setup initial location of window
x, y, w, h = 300, 200, 100, 50 # simply hardcoded the values
track_window = (x, y, w, h)
# set up the ROI for tracking
roi = frame[y:y+h, x:x+w]
hsv_roi = cv.cvtColor(roi, cv.COLOR_BGR2HSV)
mask = cv.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv.normalize(roi_hist,roi_hist,0,255,cv.NORM_MINMAX)
# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 1 )
while(1):
ret, frame = cap.read()
if ret == True:
hsv = cv.cvtColor(frame, cv.COLOR_BGR2HSV)
dst = cv.calcBackProject([hsv],[0],roi_hist,[0,180],1)
# apply camshift to get the new location
ret, track_window = cv.CamShift(dst, track_window, term_crit)
# Draw it on image
pts = cv.boxPoints(ret)
pts = np.int0(pts)
img2 = cv.polylines(frame,[pts],True, 255,2)
cv.imshow('img2',img2)
k = cv.waitKey(30) & 0xff
if k == 27:
break
else:
break

@ -0,0 +1,49 @@
import numpy as np
import cv2 as cv
import argparse
parser = argparse.ArgumentParser(description='This sample demonstrates the meanshift algorithm. \
The example file can be downloaded from: \
https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4')
parser.add_argument('image', type=str, help='path to image file')
args = parser.parse_args()
cap = cv.VideoCapture(args.image)
# take first frame of the video
ret,frame = cap.read()
# setup initial location of window
x, y, w, h = 300, 200, 100, 50 # simply hardcoded the values
track_window = (x, y, w, h)
# set up the ROI for tracking
roi = frame[y:y+h, x:x+w]
hsv_roi = cv.cvtColor(roi, cv.COLOR_BGR2HSV)
mask = cv.inRange(hsv_roi, np.array((0., 60.,32.)), np.array((180.,255.,255.)))
roi_hist = cv.calcHist([hsv_roi],[0],mask,[180],[0,180])
cv.normalize(roi_hist,roi_hist,0,255,cv.NORM_MINMAX)
# Setup the termination criteria, either 10 iteration or move by atleast 1 pt
term_crit = ( cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 1 )
while(1):
ret, frame = cap.read()
if ret == True:
hsv = cv.cvtColor(frame, cv.COLOR_BGR2HSV)
dst = cv.calcBackProject([hsv],[0],roi_hist,[0,180],1)
# apply meanshift to get the new location
ret, track_window = cv.meanShift(dst, track_window, term_crit)
# Draw it on image
x,y,w,h = track_window
img2 = cv.rectangle(frame, (x,y), (x+w,y+h), 255,2)
cv.imshow('img2',img2)
k = cv.waitKey(30) & 0xff
if k == 27:
break
else:
break

@ -0,0 +1,61 @@
import numpy as np
import cv2 as cv
import argparse
parser = argparse.ArgumentParser(description='This sample demonstrates Lucas-Kanade Optical Flow calculation. \
The example file can be downloaded from: \
https://www.bogotobogo.com/python/OpenCV_Python/images/mean_shift_tracking/slow_traffic_small.mp4')
parser.add_argument('image', type=str, help='path to image file')
args = parser.parse_args()
cap = cv.VideoCapture(args.image)
# params for ShiTomasi corner detection
feature_params = dict( maxCorners = 100,
qualityLevel = 0.3,
minDistance = 7,
blockSize = 7 )
# Parameters for lucas kanade optical flow
lk_params = dict( winSize = (15,15),
maxLevel = 2,
criteria = (cv.TERM_CRITERIA_EPS | cv.TERM_CRITERIA_COUNT, 10, 0.03))
# Create some random colors
color = np.random.randint(0,255,(100,3))
# Take first frame and find corners in it
ret, old_frame = cap.read()
old_gray = cv.cvtColor(old_frame, cv.COLOR_BGR2GRAY)
p0 = cv.goodFeaturesToTrack(old_gray, mask = None, **feature_params)
# Create a mask image for drawing purposes
mask = np.zeros_like(old_frame)
while(1):
ret,frame = cap.read()
frame_gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
# calculate optical flow
p1, st, err = cv.calcOpticalFlowPyrLK(old_gray, frame_gray, p0, None, **lk_params)
# Select good points
good_new = p1[st==1]
good_old = p0[st==1]
# draw the tracks
for i,(new,old) in enumerate(zip(good_new, good_old)):
a,b = new.ravel()
c,d = old.ravel()
mask = cv.line(mask, (a,b),(c,d), color[i].tolist(), 2)
frame = cv.circle(frame,(a,b),5,color[i].tolist(),-1)
img = cv.add(frame,mask)
cv.imshow('frame',img)
k = cv.waitKey(30) & 0xff
if k == 27:
break
# Now update the previous frame and previous points
old_gray = frame_gray.copy()
p0 = good_new.reshape(-1,1,2)

@ -0,0 +1,23 @@
import numpy as np
import cv2 as cv
cap = cv.VideoCapture(cv.samples.findFile("vtest.avi"))
ret, frame1 = cap.read()
prvs = cv.cvtColor(frame1,cv.COLOR_BGR2GRAY)
hsv = np.zeros_like(frame1)
hsv[...,1] = 255
while(1):
ret, frame2 = cap.read()
next = cv.cvtColor(frame2,cv.COLOR_BGR2GRAY)
flow = cv.calcOpticalFlowFarneback(prvs,next, None, 0.5, 3, 15, 3, 5, 1.2, 0)
mag, ang = cv.cartToPolar(flow[...,0], flow[...,1])
hsv[...,0] = ang*180/np.pi/2
hsv[...,2] = cv.normalize(mag,None,0,255,cv.NORM_MINMAX)
bgr = cv.cvtColor(hsv,cv.COLOR_HSV2BGR)
cv.imshow('frame2',bgr)
k = cv.waitKey(30) & 0xff
if k == 27:
break
elif k == ord('s'):
cv.imwrite('opticalfb.png',frame2)
cv.imwrite('opticalhsv.png',bgr)
prvs = next

@ -0,0 +1,31 @@
# Utility function: adds sample executable target with name "example_<group>_<file_name>"
# Usage:
# ocv_define_sample(<output target> <relative filename> <group>)
function(ocv_define_sample out_target source sub)
get_filename_component(name "${source}" NAME_WE)
set(the_target "example_${sub}_${name}")
add_executable(${the_target} "${source}")
if(TARGET Threads::Threads AND NOT OPENCV_EXAMPLES_DISABLE_THREADS)
target_link_libraries(${the_target} LINK_PRIVATE Threads::Threads)
endif()
set_target_properties(${the_target} PROPERTIES PROJECT_LABEL "(sample) ${name}")
if(ENABLE_SOLUTION_FOLDERS)
set_target_properties(${the_target} PROPERTIES FOLDER "samples/${sub}")
endif()
if(WIN32 AND MSVC AND NOT BUILD_SHARED_LIBS)
set_target_properties(${the_target} PROPERTIES LINK_FLAGS "/NODEFAULTLIB:atlthunk.lib /NODEFAULTLIB:atlsd.lib /DEBUG")
endif()
if(WIN32)
install(TARGETS ${the_target} RUNTIME DESTINATION "samples/${sub}" COMPONENT samples)
endif()
# Add single target to build all samples in the group: 'make opencv_samples_cpp'
set(parent_target opencv_samples_${sub})
if(NOT TARGET ${parent_target})
add_custom_target(${parent_target})
if(TARGET opencv_samples)
add_dependencies(opencv_samples ${parent_target})
endif()
endif()
add_dependencies(${parent_target} ${the_target})
set(${out_target} ${the_target} PARENT_SCOPE)
endfunction()

@ -22,5 +22,5 @@ ocv_include_modules_recurse(${OPENCV_TAPI_SAMPLES_REQUIRED_DEPS})
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} tapi)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_TAPI_SAMPLES_REQUIRED_DEPS})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_TAPI_SAMPLES_REQUIRED_DEPS})
endforeach()

@ -17,5 +17,5 @@ ocv_include_modules_recurse(${OPENCV_VA_INTEL_SAMPLES_REQUIRED_DEPS})
file(GLOB all_samples RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} *.cpp)
foreach(sample_filename ${all_samples})
ocv_define_sample(tgt ${sample_filename} va_intel)
ocv_target_link_libraries(${tgt} ${OPENCV_LINKER_LIBS} ${OPENCV_VA_INTEL_SAMPLES_REQUIRED_DEPS} ${VA_LIBRARIES} ${VA_INTEL_LIBRARIES})
ocv_target_link_libraries(${tgt} LINK_PRIVATE ${OPENCV_LINKER_LIBS} ${OPENCV_VA_INTEL_SAMPLES_REQUIRED_DEPS} ${VA_LIBRARIES} ${VA_INTEL_LIBRARIES})
endforeach()

Loading…
Cancel
Save