def recognize_from_video():
# net initialize
net = ailia.Net(MODEL_PATH, WEIGHT_PATH, env_id=args.env_id)
detector = ailia.Net(FACE_MODEL_PATH, FACE_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:
writer = webcamera_utils.get_writer(args.savepath, IMAGE_HEIGHT,
IMAGE_WIDTH)
else:
writer = None
while (True):
ret, frame = capture.read()
if (cv2.waitKey(1) & 0xFF == ord('q')) or not ret:
break
# detect face
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchors.npy',
)
# get detected face
if len(detections) == 0:
crop_img = frame
else:
crop_img, top_left, bottom_right = crop_blazeface(
detections[0], FACE_MARGIN, frame)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
crop_img = frame
# preprocess
input_image, input_data = webcamera_utils.preprocess_frame(
crop_img, IMAGE_HEIGHT, IMAGE_WIDTH, data_rgb=False)
# inference
preds_ailia = net.predict(input_data)[0]
# postprocessing
fig = gen_img_from_predsailia(input_data, preds_ailia)
fig.savefig('tmp.png')
img = cv2.imread('tmp.png')
cv2.imshow('frame', img)
# save results
if writer is not None:
img = cv2.resize(img, (IMAGE_WIDTH, IMAGE_HEIGHT))
writer.write(img)
capture.release()
cv2.destroyAllWindows()
if writer is not None:
writer.release()
os.remove('tmp.png')
print('Script finished successfully.')
def recognize_from_frame(net, detector, frame):
spoof_thresh = args.spoof_thresh
# detect face
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchorsback.npy',
back=True,
min_score_thresh=FACE_MIN_SCORE_THRESH)
# adjust face rectangle
new_detections = []
for detection in detections:
margin = 1.5
r = ailia.DetectorObject(
category=detection.category,
prob=detection.prob,
x=detection.x - detection.w * (margin - 1.0) / 2,
y=detection.y - detection.h * (margin - 1.0) / 2 -
detection.h * margin / 8,
w=detection.w * margin,
h=detection.h * margin,
)
new_detections.append(r)
# crop, preprocess
images = []
detections = []
for obj in new_detections:
# get detected face
margin = 1.0
crop_img, top_left, bottom_right = crop_blazeface(obj, margin, frame)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
continue
img = preprocess(crop_img)
images.append(img)
detections.append(
(top_left[0], top_left[1], bottom_right[0], bottom_right[1]))
if not images:
return frame
images = np.concatenate(images)
# feedforward
output = net.predict([images])
logits = output[0]
preds = softmax(logits, axis=1)
frame = draw_detections(frame, detections, preds, spoof_thresh)
return frame
def recognize_from_video():
# net initialize
net = ailia.Net(MODEL_PATH, WEIGHT_PATH, env_id=args.env_id)
if args.active_3d:
logger.info('>>> 3D mode is activated!')
depth_net = ailia.Net(DEPTH_MODEL_PATH,
DEPTH_WEIGHT_PATH,
env_id=args.env_id)
detector = ailia.Net(FACE_MODEL_PATH, FACE_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:
logger.warning('[WARNING] currently video results output feature '
'is not supported in this model!')
# TODO: shape should be debugged!
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
fig, axs = create_figure(active_3d=args.active_3d)
while (True):
ret, frame = capture.read()
if (cv2.waitKey(1) & 0xFF == ord('q')) or not ret:
break
# detect face
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchors.npy',
)
# get detected face
if len(detections) == 0:
crop_img = frame
else:
crop_img, top_left, bottom_right = crop_blazeface(
detections[0], FACE_MARGIN, frame)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
crop_img = frame
# preprocess
input_image, input_data = webcamera_utils.preprocess_frame(
crop_img, IMAGE_HEIGHT, IMAGE_WIDTH, normalize_type='255')
# inference
preds_ailia = net.predict(input_data)
pts, pts_img = get_preds_from_hm(preds_ailia)
pts, pts_img = pts.reshape(68, 2) * 4, pts_img.reshape(68, 2)
if args.active_3d:
# 3D mode
heatmaps = np.zeros((68, IMAGE_HEIGHT, IMAGE_WIDTH),
dtype=np.float32)
for i in range(68):
if pts[i, 0] > 0:
heatmaps[i] = draw_gaussian(heatmaps[i], pts[i], 2)
heatmaps = heatmaps[np.newaxis, :, :, :]
depth_pred = depth_net.predict(
np.concatenate((input_data, heatmaps), 1))
depth_pred = depth_pred.reshape(68, 1)
pts_img = np.concatenate((pts_img, depth_pred * 2), 1)
resized_img = cv2.resize(cv2.cvtColor(input_image, cv2.COLOR_BGR2RGB),
(IMAGE_WIDTH, IMAGE_HEIGHT))
# visualize results (clear axs at first)
axs = visualize_results(axs,
resized_img,
pts_img,
active_3d=args.active_3d)
plt.pause(0.01)
if not plt.get_fignums():
break
# save results
# FIXME: How to save plt --> cv2.VideoWriter()
# if writer is not None:
# # put pixel buffer in numpy array
# canvas = FigureCanvas(fig)
# canvas.draw()
# mat = np.array(canvas.renderer._renderer)
# res_img = cv2.cvtColor(mat, cv2.COLOR_RGB2BGR)
# writer.write(res_img)
capture.release()
cv2.destroyAllWindows()
if writer is not None:
writer.release()
logger.info('Script finished successfully.')
def recognize_from_video():
# net initialize
emotion_classifier = ailia.Classifier(
EMOTION_MODEL_PATH,
EMOTION_WEIGHT_PATH,
env_id=args.env_id,
format=ailia.NETWORK_IMAGE_FORMAT_GRAY,
range=ailia.NETWORK_IMAGE_RANGE_S_FP32,
channel=ailia.NETWORK_IMAGE_CHANNEL_FIRST,
)
gender_classifier = ailia.Classifier(
GENDER_MODEL_PATH,
GENDER_WEIGHT_PATH,
env_id=args.env_id,
format=ailia.NETWORK_IMAGE_FORMAT_GRAY,
range=ailia.NETWORK_IMAGE_RANGE_S_FP32,
channel=ailia.NETWORK_IMAGE_CHANNEL_FIRST,
)
detector = ailia.Net(FACE_MODEL_PATH, FACE_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 is not None:
print('[WARNING] currently video results output feature '
'is not supported in this model!')
# TODO: shape should be debugged!
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
# detect face
# WIP: FIXME: AiliaInvalidArgumentException error
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchors.npy',
)
for obj in detections:
# get detected face
crop_img, top_left, bottom_right = crop_blazeface(
obj, FACE_MARGIN, frame
)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
continue
crop_img = cv2.cvtColor(crop_img, cv2.COLOR_BGR2BGRA)
# emotion inference
emotion_classifier.compute(crop_img, EMOTION_MAX_CLASS_COUNT)
count = emotion_classifier.get_class_count()
print('=' * 80)
print(f'emotion_class_count={count}')
# print result
emotion_text = ""
for idx in range(count):
print(f'+ idx={idx}')
info = emotion_classifier.get_class(idx)
print(
f' category={info.category} ' +
f'[ {EMOTION_CATEGORY[info.category]} ]'
)
print(f' prob={info.prob}')
if idx == 0:
emotion_text = (f'[ {EMOTION_CATEGORY[info.category]} ] '
f'prob={info.prob:.3f}')
print('')
# gender inference
gender_text = ""
gender_classifier.compute(crop_img, GENDER_MAX_CLASS_COUNT)
count = gender_classifier.get_class_count()
# print reuslt
for idx in range(count):
print(f'+ idx={idx}')
info = gender_classifier.get_class(idx)
print(
f' category={info.category} ' +
f'[ {GENDER_CATEGORY[info.category]} ]'
)
print(f' prob={info.prob}')
if idx == 0:
gender_text = (f'[ {GENDER_CATEGORY[info.category]} ] '
f'prob={info.prob:.3f}')
print('')
# display label
LABEL_WIDTH = 400
LABEL_HEIGHT = 20
color = (255, 255, 255)
cv2.rectangle(frame, top_left, bottom_right, color, thickness=2)
cv2.rectangle(
frame,
top_left,
(top_left[0]+LABEL_WIDTH, top_left[1]+LABEL_HEIGHT),
color,
thickness=-1,
)
text_position = (top_left[0], top_left[1]+LABEL_HEIGHT//2)
color = (0, 0, 0)
fontScale = 0.5
cv2.putText(
frame,
emotion_text + " " + gender_text,
text_position,
cv2.FONT_HERSHEY_SIMPLEX,
fontScale,
color,
1,
)
# show result
cv2.imshow('frame', frame)
time.sleep(SLEEP_TIME)
# save results
if writer is not None:
writer.write(frame)
capture.release()
cv2.destroyAllWindows()
if writer is not None:
writer.release()
print('Script finished successfully.')
def recognize_from_frame(net, detector, frame):
# detect face
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchorsback.npy',
back=True,
min_score_thresh=FACE_MIN_SCORE_THRESH)
# adjust face rectangle
new_detections = []
for detection in detections:
margin = 1.5
r = ailia.DetectorObject(
category=detection.category,
prob=detection.prob,
x=detection.x - detection.w * (margin - 1.0) / 2,
y=detection.y - detection.h * (margin - 1.0) / 2 -
detection.h * margin / 8,
w=detection.w * margin,
h=detection.h * margin,
)
new_detections.append(r)
detections = new_detections
# estimate emotion
for obj in detections:
# get detected face
margin = 1.0
crop_img, top_left, bottom_right = crop_blazeface(obj, margin, frame)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
continue
crop_img = cv2.cvtColor(crop_img, cv2.COLOR_BGR2GRAY)
emotion = predict(net, crop_img)
idx = np.argmax(emotion)
emotion = emotion_table[idx]
# display label
LABEL_WIDTH = bottom_right[1] - top_left[1]
LABEL_HEIGHT = 20
color = (255, 128, 128)
cv2.rectangle(frame, top_left, bottom_right, color, thickness=2)
cv2.rectangle(
frame,
top_left,
(top_left[0] + LABEL_WIDTH, top_left[1] + LABEL_HEIGHT),
color,
thickness=-1,
)
text_position = (top_left[0], top_left[1] + LABEL_HEIGHT // 2)
color = (0, 0, 0)
fontScale = 0.5
cv2.putText(
frame,
emotion,
text_position,
cv2.FONT_HERSHEY_SIMPLEX,
fontScale,
color,
1,
)
def recognize_from_frame(net, detector, frame):
# detect face
detections = compute_blazeface(
detector,
frame,
anchor_path='../../face_detection/blazeface/anchorsback.npy',
back=True,
min_score_thresh=FACE_MIN_SCORE_THRESH)
# adjust face rectangle
new_detections = []
for detection in detections:
margin = 1.5
r = ailia.DetectorObject(
category=detection.category,
prob=detection.prob,
x=detection.x - detection.w * (margin - 1.0) / 2,
y=detection.y - detection.h * (margin - 1.0) / 2 -
detection.h * margin / 8,
w=detection.w * margin,
h=detection.h * margin,
)
new_detections.append(r)
detections = new_detections
# estimate age and gender
for obj in detections:
# get detected face
margin = 1.0
crop_img, top_left, bottom_right = crop_blazeface(obj, margin, frame)
if crop_img.shape[0] <= 0 or crop_img.shape[1] <= 0:
continue
img = cv2.resize(crop_img, (IMAGE_SIZE, IMAGE_SIZE))
img = np.expand_dims(img, axis=0) # 次元合せ
# inference
output = net.predict([img])
prob, age_conv3 = output
prob = prob[0][0][0]
age_conv3 = age_conv3[0][0][0][0]
i = np.argmax(prob)
gender = 'Female' if i == 0 else 'Male'
age = round(age_conv3 * 100)
# display label
LABEL_WIDTH = bottom_right[1] - top_left[1]
LABEL_HEIGHT = 20
if gender == "Male":
color = (255, 128, 128)
else:
color = (128, 128, 255)
cv2.rectangle(frame, top_left, bottom_right, color, thickness=2)
cv2.rectangle(
frame,
top_left,
(top_left[0] + LABEL_WIDTH, top_left[1] + LABEL_HEIGHT),
color,
thickness=-1,
)
text_position = (top_left[0], top_left[1] + LABEL_HEIGHT // 2)
color = (0, 0, 0)
fontScale = 0.5
cv2.putText(
frame,
"{} {}".format(gender, age),
text_position,
cv2.FONT_HERSHEY_SIMPLEX,
fontScale,
color,
1,
)