change nms before get bboxes in maximum and add check negative score,add nms for infer-det-without-torch

models/torch_utils.py

@@ -70,13 +70,13 @@ def det_postprocess(data: Tuple[Tensor, Tensor, Tensor, Tensor]):
     if nums == 0:
         return bboxes.new_zeros((0, 4)), scores.new_zeros(
             (0, )), labels.new_zeros((0, ))
+    # add nms
+    idx = nms(bboxes, scores, iou_thres)
+    bboxes, scores, labels = bboxes[idx], scores[idx], labels[idx]
     bboxes = bboxes[:nums]
     scores = scores[:nums]
     labels = labels[:nums]
 
-    # add nms
-    idx = nms(bboxes, scores, iou_thres)
-    bboxes, scores, labels = bboxes[idx], scores[idx], labels[idx]
     return bboxes, scores, labels
 
 

models/utils.py

@@ -4,6 +4,7 @@ from typing import List, Tuple, Union
 import cv2
 import numpy as np
 from numpy import ndarray
+from torchvision.ops import nms
 
 # image suffixs
 SUFFIXS = ('.bmp', '.dng', '.jpeg', '.jpg', '.mpo', '.png', '.tif', '.tiff',
@@ -87,11 +88,18 @@ def crop_mask(masks: ndarray, bboxes: ndarray) -> ndarray:
 
 def det_postprocess(data: Tuple[ndarray, ndarray, ndarray, ndarray]):
     assert len(data) == 4
+    iou_thres: float = 0.65
     num_dets, bboxes, scores, labels = (i[0] for i in data)
+    # check score negative
+    scores[scores < 0] = 1 + scores[scores < 0]
     nums = num_dets.item()
     if nums == 0:
         return np.empty((0, 4), dtype=np.float32), np.empty(
             (0, ), dtype=np.float32), np.empty((0, ), dtype=np.int32)
+    # add nms
+    idx = nms(bboxes, scores, iou_thres)
+    bboxes, scores, labels = bboxes[idx], scores[idx], labels[idx]
+
     bboxes = bboxes[:nums]
     scores = scores[:nums]
     labels = labels[:nums]