def demo_detection(net, data_dir, image_name, CLASSES, gt_roidb):
"""Detect object classes in an image using pre-computed object proposals."""
# print cfg.TEST.SCALES
# Load the demo image
im1_file = os.path.join(data_dir, 'color', 'I' + image_name + '.jpg')
im2_file = os.path.join(data_dir, 'thermal', 'I' + image_name + '.jpg')
print im1_file
print im2_file
im1 = cv2.imread(im1_file)
im2 = cv2.imread(im2_file)
# Detect all object classes and regress object bounds
timer = Timer()
timer.tic()
scores, boxes = im_detect_2in(net, im1, im2)
timer.toc()
print('Detection took {:.3f}s for '
'{:d} object proposals').format(timer.total_time, boxes.shape[0])
# Visualize detections for each class
#CONF_THRESH = 0.1
NMS_THRESH = 0.3
# NMS_THRESH = [0.2,0.2, 0.2]
all_dets = None
#for cls_ind, cls in enumerate(classes):
# cls_ind += 1 # because we skipped background
for cls_ind, cls in enumerate(CLASSES[1:]):
cls_ind += 1
cls_boxes = boxes[:, 4 * cls_ind:4 * (cls_ind + 1)]
cls_scores = scores[:, cls_ind]
dets = np.hstack(
(cls_boxes, cls_scores[:, np.newaxis])).astype(np.float32)
keep = nms(dets, NMS_THRESH)
dets = dets[keep, :]
cls_inds = np.ones((len(keep), )) * cls_ind
#cls_inds = np.ones((len(cls_scores),)) * cls_ind
#cls_inds = np.ones((len(cls_scores),)) * (-1)
#cls_inds[keep] = cls_ind
dets = np.hstack((cls_inds[:, np.newaxis], dets))
if all_dets is None:
all_dets = dets
else:
all_dets = np.vstack((all_dets, dets))
visual_detection_results(im1, boxes, scores, CLASSES, gt_roidb, threds=0.5)
return all_dets
def detect(qf, qex, qdet, qtrack, qbox, qnum, f, isInited):
caffe.set_mode_gpu()
caffe.set_device(0)
# f=1
# a=f
net = caffe.Net(prototxt, caffemodel, caffe.TEST)
data_dir = os.path.join(cfg.ROOT_DIR, 'data/demo_pedestrian/set07_V000')
CLASSES = ('__background__', 'person')
isInited.value = True
while True:
im = qdet.get()
if (im == 'done'):
break
im = format(im, '05d')
qex.put(im)
# print(im)
im1_file = os.path.join(data_dir, 'color', 'I' + im + '.jpg')
im2_file = os.path.join(data_dir, 'thermal', 'I' + im + '.jpg')
# print (im1_file)
# print (im2_file)
im1 = cv2.imread(im1_file)
im2 = cv2.imread(im2_file)
timer = Timer()
timer.tic()
scores, boxes = im_detect_2in(net, im1, im2)
#f.value = False
#fps.update()
timer.toc()
# print("1111111111111111111111111111111111111111111")
# print (f'Detection took {timer.total_time}s')
NMS_THRESH = 0.3
all_dets = None
# cls_boxes = boxes[:, 4:4 * (cls_ind + 1)]
box = []
for cls_ind, cls in enumerate(CLASSES[1:]):
cls_ind += 1
cls_boxes = boxes[:, 4 * cls_ind:4 * (cls_ind + 1)]
cls_scores = scores[:, cls_ind]
dets = np.hstack(
(cls_boxes, cls_scores[:, np.newaxis])).astype(np.float32)
keep = nms(dets, NMS_THRESH)
dets = dets[keep, :]
cls_inds = np.ones((len(keep), )) * cls_ind
inds = np.where(dets[:, -1] > 0.5)[0]
for i in inds:
bbox = dets[i, :4]
score = dets[i, -1]
if (bbox[0] < 0):
bbox[0] = 0
if (bbox[1] < 0):
bbox[1] = 0
bb = (bbox[0], bbox[1], bbox[2] - bbox[0], bbox[3] - bbox[1])
box.append(bb)
# ax.add_patch(
# plt.Rectangle((bbox[0], bbox[1]),
# bbox[2] - bbox[0],
# bbox[3] - bbox[1], fill=False,
# edgecolor=cls_color[2], linewidth=3.5)
# )
# ax.text(bbox[0], bbox[1] - 2,
# '{:s} {:.3f}'.format(cls, score),
# bbox=dict(facecolor='blue', alpha=0.5),
# fontsize=14, color='white')
# qnum.put(inds)
# if (len(box)>0):
qbox.put(box)
f.value = False
#print("Detet: ",str(qnum.qsize()))
# time.sleep(5)
#print(1111)
# qf.get()
# print("f235556 is: ", str(qf.qsize()))
print('out o det')
qtrack.put('done')