You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 

221 lines
8.2 KiB

# Ultralytics YOLO 🚀, AGPL-3.0 license
import argparse
from typing import Tuple, Union
import cv2
import numpy as np
import tensorflow as tf
import yaml
from ultralytics.utils import ASSETS
try:
from tflite_runtime.interpreter import Interpreter
except ImportError:
import tensorflow as tf
Interpreter = tf.lite.Interpreter
class YOLOv8TFLite:
"""
YOLOv8TFLite.
A class for performing object detection using the YOLOv8 model with TensorFlow Lite.
Attributes:
model (str): Path to the TensorFlow Lite model file.
conf (float): Confidence threshold for filtering detections.
iou (float): Intersection over Union threshold for non-maximum suppression.
metadata (Optional[str]): Path to the metadata file, if any.
Methods:
detect(img_path: str) -> np.ndarray:
Performs inference and returns the output image with drawn detections.
"""
def __init__(self, model: str, conf: float = 0.25, iou: float = 0.45, metadata: Union[str, None] = None):
"""
Initializes an instance of the YOLOv8TFLite class.
Args:
model (str): Path to the TFLite model.
conf (float, optional): Confidence threshold for filtering detections. Defaults to 0.25.
iou (float, optional): IoU (Intersection over Union) threshold for non-maximum suppression. Defaults to 0.45.
metadata (Union[str, None], optional): Path to the metadata file or None if not used. Defaults to None.
"""
self.conf = conf
self.iou = iou
if metadata is None:
self.classes = {i: i for i in range(1000)}
else:
with open(metadata) as f:
self.classes = yaml.safe_load(f)["names"]
np.random.seed(42)
self.color_palette = np.random.uniform(128, 255, size=(len(self.classes), 3))
self.model = Interpreter(model_path=model)
self.model.allocate_tensors()
input_details = self.model.get_input_details()[0]
self.in_width, self.in_height = input_details["shape"][1:3]
self.in_index = input_details["index"]
self.in_scale, self.in_zero_point = input_details["quantization"]
self.int8 = input_details["dtype"] == np.int8
output_details = self.model.get_output_details()[0]
self.out_index = output_details["index"]
self.out_scale, self.out_zero_point = output_details["quantization"]
def letterbox(self, img: np.ndarray, new_shape: Tuple = (640, 640)) -> Tuple[np.ndarray, Tuple[float, float]]:
"""Resizes and reshapes images while maintaining aspect ratio by adding padding, suitable for YOLO models."""
shape = img.shape[:2] # current shape [height, width]
# Scale ratio (new / old)
r = min(new_shape[0] / shape[0], new_shape[1] / shape[1])
# Compute padding
new_unpad = int(round(shape[1] * r)), int(round(shape[0] * r))
dw, dh = (new_shape[1] - new_unpad[0]) / 2, (new_shape[0] - new_unpad[1]) / 2 # wh padding
if shape[::-1] != new_unpad: # resize
img = cv2.resize(img, new_unpad, interpolation=cv2.INTER_LINEAR)
top, bottom = int(round(dh - 0.1)), int(round(dh + 0.1))
left, right = int(round(dw - 0.1)), int(round(dw + 0.1))
img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=(114, 114, 114))
return img, (top / img.shape[0], left / img.shape[1])
def draw_detections(self, img: np.ndarray, box: np.ndarray, score: np.float32, class_id: int) -> None:
"""
Draws bounding boxes and labels on the input image based on the detected objects.
Args:
img (np.ndarray): The input image to draw detections on.
box (np.ndarray): Detected bounding box in the format [x1, y1, width, height].
score (np.float32): Corresponding detection score.
class_id (int): Class ID for the detected object.
Returns:
None
"""
x1, y1, w, h = box
color = self.color_palette[class_id]
cv2.rectangle(img, (int(x1), int(y1)), (int(x1 + w), int(y1 + h)), color, 2)
label = f"{self.classes[class_id]}: {score:.2f}"
(label_width, label_height), _ = cv2.getTextSize(label, cv2.FONT_HERSHEY_SIMPLEX, 0.5, 1)
label_x = x1
label_y = y1 - 10 if y1 - 10 > label_height else y1 + 10
cv2.rectangle(
img,
(int(label_x), int(label_y - label_height)),
(int(label_x + label_width), int(label_y + label_height)),
color,
cv2.FILLED,
)
cv2.putText(img, label, (int(label_x), int(label_y)), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 1, cv2.LINE_AA)
def preprocess(self, img: np.ndarray) -> Tuple[np.ndarray, Tuple[float, float]]:
"""
Preprocesses the input image before performing inference.
Args:
img (np.ndarray): The input image to be preprocessed.
Returns:
Tuple[np.ndarray, Tuple[float, float]]: A tuple containing:
- The preprocessed image (np.ndarray).
- A tuple of two float values representing the padding applied (top/bottom, left/right).
"""
img, pad = self.letterbox(img, (self.in_width, self.in_height))
img = img[..., ::-1][None] # N,H,W,C for TFLite
img = np.ascontiguousarray(img)
img = img.astype(np.float32)
return img / 255, pad
def postprocess(self, img: np.ndarray, outputs: np.ndarray, pad: Tuple[float, float]) -> np.ndarray:
"""
Performs post-processing on the model's output to extract bounding boxes, scores, and class IDs.
Args:
img (numpy.ndarray): The input image.
outputs (numpy.ndarray): The output of the model.
pad (Tuple[float, float]): Padding used by letterbox.
Returns:
numpy.ndarray: The input image with detections drawn on it.
"""
outputs[:, 0] -= pad[1]
outputs[:, 1] -= pad[0]
outputs[:, :4] *= max(img.shape)
outputs = outputs.transpose(0, 2, 1)
outputs[..., 0] -= outputs[..., 2] / 2
outputs[..., 1] -= outputs[..., 3] / 2
for out in outputs:
scores = out[:, 4:].max(-1)
keep = scores > self.conf
boxes = out[keep, :4]
scores = scores[keep]
class_ids = out[keep, 4:].argmax(-1)
indices = cv2.dnn.NMSBoxes(boxes, scores, self.conf, self.iou).flatten()
[self.draw_detections(img, boxes[i], scores[i], class_ids[i]) for i in indices]
return img
def detect(self, img_path: str) -> np.ndarray:
"""
Performs inference using a TFLite model and returns the output image with drawn detections.
Args:
img_path (str): The path to the input image file.
Returns:
np.ndarray: The output image with drawn detections.
"""
img = cv2.imread(img_path)
x, pad = self.preprocess(img)
if self.int8:
x = (x / self.in_scale + self.in_zero_point).astype(np.int8)
self.model.set_tensor(self.in_index, x)
self.model.invoke()
y = self.model.get_tensor(self.out_index)
if self.int8:
y = (y.astype(np.float32) - self.out_zero_point) * self.out_scale
return self.postprocess(img, y, pad)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument(
"--model",
type=str,
default="yolov8n_saved_model/yolov8n_full_integer_quant.tflite",
help="Path to TFLite model.",
)
parser.add_argument("--img", type=str, default=str(ASSETS / "bus.jpg"), help="Path to input image")
parser.add_argument("--conf", type=float, default=0.25, help="Confidence threshold")
parser.add_argument("--iou", type=float, default=0.45, help="NMS IoU threshold")
parser.add_argument("--metadata", type=str, default="yolov8n_saved_model/metadata.yaml", help="Metadata yaml")
args = parser.parse_args()
detector = YOLOv8TFLite(args.model, args.conf, args.iou, args.metadata)
result = detector.detect(str(ASSETS / "bus.jpg"))[..., ::-1]
cv2.imshow("Output", result)
cv2.waitKey(0)