def recognize_from_image():
# prepare input data
org_img = load_image(
args.input,
(IMAGE_HEIGHT, IMAGE_WIDTH),
)
input_data = load_image(args.input, (IMAGE_HEIGHT, IMAGE_WIDTH),
normalize_type='127.5',
gen_input_ailia=True)
# net initialize
net = ailia.Net(MODEL_PATH, WEIGHT_PATH, env_id=args.env_id)
# inference
print('Start inference...')
if args.benchmark:
print('BENCHMARK mode')
for i in range(5):
start = int(round(time.time() * 1000))
preds_ailia = net.predict([input_data])
end = int(round(time.time() * 1000))
print(f'\tailia processing time {end - start} ms')
else:
preds_ailia = net.predict([input_data])
# postprocessing
detections = but.postprocess(preds_ailia)
# generate detections
for detection in detections:
but.plot_detections(org_img, detection, save_image_path=args.savepath)
print('Script finished successfully.')
def recognize_from_video(net):
if args.back == True:
IMAGE_HEIGHT = IMAGE_HEIGHT_BACK
IMAGE_WIDTH = IMAGE_WIDTH_BACK
ANCHOR_PATH = ANCHOR_PATH_BACK
else:
IMAGE_HEIGHT = IMAGE_HEIGHT_FRONT
IMAGE_WIDTH = IMAGE_WIDTH_FRONT
ANCHOR_PATH = ANCHOR_PATH_FRONT
capture = webcamera_utils.get_capture(args.video)
# create video writer if savepath is specified as video format
if args.savepath != SAVE_IMAGE_PATH:
f_h = int(capture.get(cv2.CAP_PROP_FRAME_HEIGHT))
f_w = int(capture.get(cv2.CAP_PROP_FRAME_WIDTH))
writer = webcamera_utils.get_writer(args.savepath, f_h, f_w)
else:
writer = None
while (True):
ret, frame = capture.read()
if (cv2.waitKey(1) & 0xFF == ord('q')) or not ret:
break
input_image, input_data = webcamera_utils.preprocess_frame(
frame, IMAGE_HEIGHT, IMAGE_WIDTH, normalize_type='127.5')
# inference
input_blobs = net.get_input_blob_list()
net.set_input_blob_data(input_data, input_blobs[0])
net.update()
preds_ailia = net.get_results()
# postprocessing
detections = but.postprocess(preds_ailia,
anchor_path=ANCHOR_PATH,
back=args.back)
but.show_result(input_image, detections)
# remove padding
dh = input_image.shape[0]
dw = input_image.shape[1]
sh = frame.shape[0]
sw = frame.shape[1]
input_image = input_image[(dh - sh) // 2:(dh - sh) // 2 + sh,
(dw - sw) // 2:(dw - sw) // 2 + sw, :]
cv2.imshow('frame', input_image)
# save results
if writer is not None:
writer.write(input_image)
capture.release()
cv2.destroyAllWindows()
if writer is not None:
writer.release()
logger.info('Script finished successfully.')
def face_detect(img, face_net):
IMAGE_BLAZE_SIZE = 128
img_0 = img
img = normalize_image(img, normalize_type='127.5')
img = cv2.resize(img, (IMAGE_BLAZE_SIZE, IMAGE_BLAZE_SIZE))
img = img.transpose((2, 0, 1))
img = np.expand_dims(img, axis=0)
output = face_net.predict([img])
detections = but.postprocess(output)
detections = detections[0]
# sort by confidence
detections = sorted(detections, key=lambda x: x[16], reverse=True)
if len(detections) == 0:
return None, (0, 0)
detection = detections[0]
h, w = img_0.shape[:2]
ymin = int(detection[0] * h)
xmin = int(detection[1] * w)
ymax = int(detection[2] * h)
xmax = int(detection[3] * w)
h = ymax - ymin
w = xmax - xmin
if h > w:
p = (h - w) // 2
w = h
xmin -= p
else:
p = (w - h) // 2
h = w
ymin -= p
img = img_0[ymin:ymin + h, xmin:xmin + w]
h2, w2 = img.shape[:2]
if h != h2 or w != w2:
return None, (0, 0)
return img, (ymin, xmin)
def recognize_from_video():
# net initialize
net = ailia.Net(MODEL_PATH, WEIGHT_PATH, env_id=args.env_id)
capture = webcamera_utils.get_capture(args.video)
# create video writer if savepath is specified as video format
if args.savepath != SAVE_IMAGE_PATH:
f_h = int(capture.get(cv2.CAP_PROP_FRAME_HEIGHT))
f_w = int(capture.get(cv2.CAP_PROP_FRAME_WIDTH))
save_h, save_w = webcamera_utils.calc_adjust_fsize(
f_h, f_w, IMAGE_HEIGHT, IMAGE_WIDTH)
writer = webcamera_utils.get_writer(args.savepath, save_h, save_w)
else:
writer = None
while (True):
ret, frame = capture.read()
if (cv2.waitKey(1) & 0xFF == ord('q')) or not ret:
break
input_image, input_data = webcamera_utils.preprocess_frame(
frame, IMAGE_HEIGHT, IMAGE_WIDTH, normalize_type='127.5')
# inference
input_blobs = net.get_input_blob_list()
net.set_input_blob_data(input_data, input_blobs[0])
net.update()
preds_ailia = net.get_results()
# postprocessing
detections = but.postprocess(preds_ailia)
but.show_result(input_image, detections)
cv2.imshow('frame', input_image)
# save results
if writer is not None:
writer.write(input_image)
capture.release()
cv2.destroyAllWindows()
if writer is not None:
writer.release()
print('Script finished successfully.')
def recognize_from_image():
# net initialize
net = ailia.Net(MODEL_PATH, WEIGHT_PATH, env_id=args.env_id)
# input image loop
for image_path in args.input:
# prepare input data
logger.info(image_path)
org_img = load_image(image_path, (IMAGE_HEIGHT, IMAGE_WIDTH))
input_data = load_image(image_path, (IMAGE_HEIGHT, IMAGE_WIDTH),
normalize_type='127.5',
gen_input_ailia=True)
# inference
logger.info('Start inference...')
if args.benchmark:
logger.info('BENCHMARK mode')
total_time = 0
for i in range(args.benchmark_count):
start = int(round(time.time() * 1000))
preds_ailia = net.predict([input_data])
end = int(round(time.time() * 1000))
logger.info(f'\tailia processing time {end - start} ms')
if i != 0:
total_time = total_time + (end - start)
logger.info(
f'\taverage time {total_time / (args.benchmark_count-1)} ms')
else:
preds_ailia = net.predict([input_data])
# post-processing
detections = but.postprocess(preds_ailia)
# generate detections
savepath = get_savepath(args.savepath, image_path)
logger.info(f'saved at : {savepath}')
for detection in detections:
but.plot_detections(org_img, detection, save_image_path=savepath)
logger.info('Script finished successfully.')
def recognize_from_image(net):
if args.back == True:
IMAGE_HEIGHT = IMAGE_HEIGHT_BACK
IMAGE_WIDTH = IMAGE_WIDTH_BACK
ANCHOR_PATH = ANCHOR_PATH_BACK
else:
IMAGE_HEIGHT = IMAGE_HEIGHT_FRONT
IMAGE_WIDTH = IMAGE_WIDTH_FRONT
ANCHOR_PATH = ANCHOR_PATH_FRONT
# input image loop
for image_path in args.input:
# prepare input data
logger.info(image_path)
org_img = cv2.imread(image_path, cv2.IMREAD_COLOR)
input_data = load_image(image_path)
input_data = cv2.cvtColor(input_data, cv2.COLOR_BGRA2RGB)
input_data, pad_hw, resized_hw = preprocess(
input_data, (IMAGE_HEIGHT, IMAGE_WIDTH))
# inference
logger.info('Start inference...')
if args.benchmark:
logger.info('BENCHMARK mode')
total_time = 0
for i in range(args.benchmark_count):
start = int(round(time.time() * 1000))
preds_ailia = net.predict([input_data])
end = int(round(time.time() * 1000))
logger.info(f'\tailia processing time {end - start} ms')
if i != 0:
total_time = total_time + (end - start)
logger.info(
f'\taverage time {total_time / (args.benchmark_count - 1)} ms')
else:
preds_ailia = net.predict([input_data])
# post-processing
detections = but.postprocess(preds_ailia,
anchor_path=ANCHOR_PATH,
back=args.back)
# remove padding
pad_x = pad_hw[1] / IMAGE_WIDTH
pad_y = pad_hw[0] / IMAGE_HEIGHT
resized_x = resized_hw[1] / IMAGE_WIDTH
resized_y = resized_hw[0] / IMAGE_HEIGHT
for d in detections:
d[:, [1, 3, 4, 6, 8, 10, 12, 14]] = (
d[:, [1, 3, 4, 6, 8, 10, 12, 14]] - pad_x) / resized_x
d[:, [0, 2, 5, 7, 9, 11, 13, 15]] = (
d[:, [0, 2, 5, 7, 9, 11, 13, 15]] - pad_y) / resized_y
# generate detections
savepath = get_savepath(args.savepath, image_path)
logger.info(f'saved at : {savepath}')
for detection in detections:
but.plot_detections(org_img, detection, save_image_path=savepath)
logger.info('Script finished successfully.')