def detect(image, model, num_classes, max_per_image):
images, meta = pre_process(image, 1)
images = images.to("cuda")
output, dets = process(images, model, return_time=True)
detection_result = []
dets = post_process(dets, meta, num_classes, max_per_image)
results = merge_outputs(dets, num_classes, max_per_image)
images = images.to("cpu")
image_detection = None
# image_detection = np.zeros((384,384,3))
for j in range(1, num_classes + 1):
for bbox in results[j]:
# print("the bbox is {}".format(bbox))
# print("the type of bbox is ",type(bbox))
if bbox[4] > 0.01:
detection_result.append(
[bbox[0], bbox[1], bbox[2], bbox[3], bbox[4], j])
image_detection = add_coco_bbox(image,
bbox,
j - 1,
conf=1,
show_txt=True,
img_id='default')
# else:
# detection_result.append([0,0,0,0,0])
# cv2.imshow("detection",image_detection)
# cv2.waitKey(10)
return image_detection, detection_result
def detect_eval(image, model, num_classes, max_per_image):
images, meta = pre_process(image, 1)
images = images.to("cuda")
output, dets = process(images, model, return_time=True)
detection_result = []
dets = post_process(dets, meta, num_classes, max_per_image)
results = merge_outputs(dets, num_classes, max_per_image)
# images = images.to("cpu")
for j in range(1, num_classes + 1):
for bbox in results[j]:
if bbox[4] > 0.1:
detection_result.append(
[bbox[0], bbox[1], bbox[2], bbox[3], bbox[4], j - 1])
return detection_result
def detect(image, model, num_classes):
images, meta = pre_process(image, 1)
images = images.to("cuda")
output, dets = process(images, model, return_time=True)
# print("the wh is {}".format(output["wh"]))
dets = post_process(dets, meta, num_classes)
results = merge_outputs(dets, num_classes, max_per_image)
for j in range(1, num_classes + 1):
for bbox in results[j]:
if bbox[4] > 0.3:
image_detection = add_coco_bbox(image,
bbox,
bbox[4],
conf=1,
show_txt=True,
img_id='default')
cv2.imshow("detection", image_detection)
cv2.waitKey(1)
return image_detection