def show_detections(self, head_pose_detections, frame):
for det in head_pose_detections:
utils.draw_axis(frame,
det.yaw,
det.pitch,
det.roll,
tdx=(det.x_min + det.x_max) / 2,
tdy=(det.y_min + det.y_max) / 2,
size=det.bbox_height / 2)
cv2.rectangle(frame, (det.x_min, det.y_min),
(det.x_max, det.y_max), (0, 255, 0), 1)
def save_test(self, name, save_dir, prediction):
img_path = os.path.join(self.data_dir, name + '.png')
print("IMG PATH:", img_path)
cv2_img = cv2.imread(img_path)
print("PREDICT:", prediction)
cv2_img = utils.draw_axis(cv2_img, prediction[0], prediction[1], prediction[2],
size=50)
save_path = os.path.join(save_dir, name + '.png')
print("SAVE PATH:", save_path)
cv2.imwrite(save_path, cv2_img)
def draw_detection(img, bbox, yaw, pitch, roll):
y_min, x_min, y_max, x_max = bbox
y_min = max(0, y_min - abs(y_min - y_max) / 10)
y_max = min(img.shape[0], y_max + abs(y_min - y_max) / 10)
x_min = max(0, x_min - abs(x_min - x_max) / 5)
x_max = min(img.shape[1], x_max + abs(x_min - x_max) / 5)
x_max = min(x_max, img.shape[1])
draw_axis(img,
yaw,
pitch,
roll,
tdx=(x_min + x_max) / 2,
tdy=(y_min + y_max) / 2,
size=abs(x_max - x_min) // 2)
cv2.putText(img, "yaw: {}".format(np.round(yaw)), (int(x_min), int(y_min)),
cv2.FONT_HERSHEY_SIMPLEX, 0.4, (100, 255, 0), 1)
cv2.putText(img, "pitch: {}".format(np.round(pitch)),
(int(x_min), int(y_min) - 15), cv2.FONT_HERSHEY_SIMPLEX, 0.4,
(100, 255, 0), 1)
cv2.putText(img, "roll: {}".format(np.round(roll)),
(int(x_min), int(y_min) - 30), cv2.FONT_HERSHEY_SIMPLEX, 0.4,
(100, 255, 0), 1)
def save_result(self, name, yaw, pitch, roll):
image = cv2.imread(os.path.join(self.test_dir, name + self.test_format))
cv2_img = utils.draw_axis(image, yaw, pitch, roll, tdx=image.shape[0]//2, tdy=image.shape[1]//2,size=100)
save_path = os.path.join(self.result_dir, name + self.test_format)
cv2.imwrite(save_path, cv2_img)
draw = cv2.rectangle(draw,
(face_boxes[i][0], face_boxes[i][1]),
(face_boxes[i][2], face_boxes[i][3]),
(0, 0, 255), 2)
face_box_width = face_boxes[i][2] - face_boxes[i][0]
axis_x, axis_y = utils.unnormalize_landmark_point(
(landmark[4], landmark[5]),
net_input_size,
scale=scale)
axis_x += face_boxes[i][0]
axis_y += face_boxes[i][1]
draw = utils.draw_axis(
draw,
yaw,
pitch,
roll,
tdx=face_boxes[i][0] + face_box_width // 5,
tdy=face_boxes[i][1] + face_box_width // 5,
size=face_box_width // 2)
for j in range(5):
x = landmark[2 * j]
y = landmark[2 * j + 1]
x, y = utils.unnormalize_landmark_point((x, y),
net_input_size,
scale=scale)
x += face_boxes[i][0]
y += face_boxes[i][1]
x = int(x)
y = int(y)
draw = cv2.circle(draw, (x, y), 2, (255, 0, 0), 2)
pitch_predicted = softmax(pitch)
roll_predicted = softmax(roll)
# Get continuous predictions in degrees.
yaw_predicted = torch.sum(
yaw_predicted.data[0] * idx_tensor) * 3 - 99
pitch_predicted = torch.sum(
pitch_predicted.data[0] * idx_tensor) * 3 - 99
roll_predicted = torch.sum(
roll_predicted.data[0] * idx_tensor) * 3 - 99
# Print new frame with cube and axis
# utils.plot_pose_cube(frame, yaw_predicted, pitch_predicted, roll_predicted, (xw1 + xw2) / 2, (yw1 + yw2) / 2, size = w)
utils.draw_axis(frame,
yaw_predicted,
pitch_predicted,
roll_predicted,
tdx=(xw1 + xw2) / 2,
tdy=(yw1 + yw2) / 2,
size=w / 2)
bounding_box = d['box']
cv2.rectangle(frame,(bounding_box[0], bounding_box[1]), \
(bounding_box[0]+bounding_box[2], bounding_box[1] + bounding_box[3]), \
(140,255,255), 3)
out.write(frame)
frame_num += 1
out.release()
video.release()