def get_more_data(img, outputs, class_names):
boxes, objectness, classes, nums = outputs
boxes, objectness, classes, nums = boxes[0], objectness[0], classes[
0], nums[0]
wh = np.flip(img.shape[0:2])
process_s = []
nb_image = 0
for i in range(nums):
x1y1 = tuple((np.array(boxes[i][0:2]) * wh).astype(np.int32))
x2y2 = tuple((np.array(boxes[i][2:4]) * wh).astype(np.int32))
if class_names[int(classes[i])] == "cat" or class_names[int(
classes[i])] == "dog":
x1, y1 = x1y1
x2, y2 = x2y2
img_crop = img[y1:y1 + (y2 - y1), x1:x1 + (x2 - x1)]
nb_image = nb_image + 1
img_to_net = img_t.load_img_opencv(img_crop)
if class_names[int(classes[i])] == "cat":
process = ThreadSend("cat", img_to_net)
process.setDaemon(True)
process.start()
process_s.append(process)
else:
process = ThreadSend("dog", img_to_net)
process.setDaemon(True)
process.start()
process_s.append(process)
return nb_image
def draw_outputs(img, outputs, class_names):
boxes, objectness, classes, nums = outputs
boxes, objectness, classes, nums = boxes[0], objectness[0], classes[
0], nums[0]
wh = np.flip(img.shape[0:2])
dog_det = []
cat_det = []
for i in range(nums):
x1y1 = tuple((np.array(boxes[i][0:2]) * wh).astype(np.int32))
x2y2 = tuple((np.array(boxes[i][2:4]) * wh).astype(np.int32))
x1, y1 = x1y1
x2, y2 = x2y2
img_crop = img[y1:y1 + (y2 - y1), x1:x1 + (x2 - x1)]
img_to_cnn = img_t.load_img_opencv(img_crop)
label = ""
prob = 0
if class_names[int(classes[i])] == "dog":
label, prob = cnn_t.get_more_dog(img_to_cnn)
dog_det.append([prob, label])
if class_names[int(classes[i])] == "cat":
prob, label = cnn_t.get_more_cat(img_to_cnn)
cat_det.append([label, prob])
img = cv2.rectangle(img, x1y1, x2y2, (0, 0, 0), 2)
img = cv2.putText(
img, '{} {:.2f} {} {:.2f}'.format(class_names[int(classes[i])],
objectness[i], label, prob),
x1y1, cv2.FONT_HERSHEY_COMPLEX_SMALL, 1, (255, 255, 255), 2)
return img, cat_det, dog_det
def get_more_data(img, model_cat, model_dog, outputs, class_names):
boxes, objectness, classes, nums = outputs
boxes, objectness, classes, nums = boxes[0], objectness[0], classes[
0], nums[0]
wh = np.flip(img.shape[0:2])
process_s = []
for i in range(nums):
x1y1 = tuple((np.array(boxes[i][0:2]) * wh).astype(np.int32))
x2y2 = tuple((np.array(boxes[i][2:4]) * wh).astype(np.int32))
if class_names[int(classes[i])] == "cat" or class_names[int(
classes[i])] == "dog":
x1, y1 = x1y1
x2, y2 = x2y2
img_crop = img[y1:y1 + (y2 - y1), x1:x1 + (x2 - x1)]
img_to_cnn = img_t.load_img_opencv(img_crop)
if class_names[int(classes[i])] == "cat":
process = ThreadCat(i, "cat", model_cat, img_to_cnn)
process_s.append(process)
elif class_names[int(classes[i])] == "dog":
process = ThreadDog(i, "dog", model_dog, img_to_cnn)
process_s.append(process)
for process in process_s:
process.start()
process.join()
return info_cat, info_dog, info_box
def thread_for_cnn(i, classe, model, img):
img_to_cnn = img_t.load_img_opencv(img)
label = ""
prob = 0
if classe == "dog":
label, prob = get_more_dog(model, img_to_cnn)
else:
label, prob = get_more_cat(model, img_to_cnn)
info.append([i, label, prob])
def send_data(cat, data, s):
data = img_tools.load_img_opencv(data)
print("Type : ", type(data), "Dimension : ", np.shape(data))
message = "1;" + "YOLO" + ";" + cat.upper() + ";" + np.array2string(
data) + ";"
logging.info("Send image category : " + cat + " message size = " +
str(sys.getsizeof(message)))
s.send(message.encode('utf-8'))