def load_info_from_json(json_path):
data = gnu.load_json(json_path)
# image path
assert ('image_path' in data), "Path of input image should be specified"
image_path = data['image_path']
assert osp.exists(image_path), f"{image_path} does not exists"
# body bboxes
body_bbox_list = list()
if 'body_bbox_list' in data:
body_bbox_list = data['body_bbox_list']
assert isinstance(body_bbox_list, list)
for b_id, body_bbox in enumerate(body_bbox_list):
if isinstance(body_bbox, list) and len(body_bbox) == 4:
body_bbox_list[b_id] = np.array(body_bbox)
# hand bboxes
hand_bbox_list = list()
if 'hand_bbox_list' in data:
hand_bbox_list = data['hand_bbox_list']
assert isinstance(hand_bbox_list, list)
for hand_bbox in hand_bbox_list:
for hand_type in ['left_hand', 'right_hand']:
if hand_type in hand_bbox:
bbox = hand_bbox[hand_type]
if isinstance(bbox, list) and len(bbox) == 4:
hand_bbox[hand_type] = np.array(bbox)
else:
hand_bbox[hand_type] = None
return image_path, body_bbox_list, hand_bbox_list
def run_hand_mocap(args, bbox_detector, hand_mocap, visualizer):
#Set up input data (images or webcam)
input_type, input_data = demo_utils.setup_input(args)
assert args.out_dir is not None, "Please specify output dir to store the results"
video_frame = 0
#pbar = tqdm(total=13000)
while True:
# load data
load_bbox = False
if input_type == 'video':
_, img_original_bgr = input_data.read()
if video_frame < args.start_frame:
video_frame += 1
continue
# save the obtained video frames
image_path = osp.join(args.out_dir, "frames",
f"{video_frame:05d}.jpg")
else:
assert False, "Unknown input_type"
if img_original_bgr is None or video_frame > args.end_frame:
break
#print("--------------------------------------")
# bbox detection
if args.load_bboxes:
#print(osp.join(args.out_dir, WHICH_BBOXES))
if not osp.exists(osp.join(args.out_dir, WHICH_BBOXES)):
print(f"[ERROR] {WHICH_BBOXES} folder do not exist.")
return -1
try:
json_name = f"{video_frame:05d}_bbox.json"
json_path = osp.join(args.out_dir, WHICH_BBOXES, json_name)
data = gnu.load_json(json_path)
#print(f"Data loaded from '{json_path}'")
#[{'left_hand': array([608.8389 , 559.77545, 111.33954, 97.50702], dtype=float32), 'right_hand': array([463.8986 , 550.34717, 89.06082, 93.86871], dtype=float32), 'left_hand_score': 0.99971753, 'right_hand_score': 0.99974126}]
hands = {}
if data["left_hand"] is not None:
hands["left_hand"] = np.array(data["left_hand"]).astype(
np.float32)
if data["right_hand"] is not None:
hands["right_hand"] = np.array(data["right_hand"]).astype(
np.float32)
hand_bbox_list = [
hands,
]
body_bbox_list = []
if data["body"] is not None:
body_bbox_list.append(
np.array(data["body"]).astype(np.float32))
except:
video_frame += 1
print(f"[WARNING_ANNOTATIONS] Frame {video_frame}")
continue
else:
raise Exception("Should have computed already the bboxes, yeah?")
if len(hand_bbox_list) < 1:
print(f"No hand detected: {image_path}")
video_frame += 1
continue
# Hand Pose Regression
pred_output_list = hand_mocap.regress(img_original_bgr,
hand_bbox_list,
add_margin=True)
assert len(hand_bbox_list) == len(body_bbox_list)
assert len(body_bbox_list) == len(pred_output_list)
# save predictions to pkl
if args.save_pred_pkl:
demo_type = 'hand'
demo_utils.save_pred_to_pkl(args, demo_type, image_path,
body_bbox_list, hand_bbox_list,
pred_output_list)
#print(f"Processed : {image_path}")
video_frame += 1
#pbar.update(1)
with open(os.path.join(args.out_dir, MARKER), "w+") as f:
f.write("True")
def run_body_mocap(args, body_bbox_detector, body_mocap, visualizer):
#Setup input data to handle different types of inputs
input_type, input_data = demo_utils.setup_input(args)
cur_frame = args.start_frame
video_frame = 0
timer = Timer()
while True:
timer.tic()
# always video
_, img_original_bgr = input_data.read()
if video_frame < cur_frame:
video_frame += 1
continue
# save the obtained video frames
if img_original_bgr is not None:
video_frame += 1
cur_frame +=1
if img_original_bgr is None or cur_frame > args.end_frame:
break
print("--------------------------------------")
# save the obtained body & hand bbox to json file
if args.load_bboxes:
json_name = f"{cur_frame:05d}_bbox.json"
json_path = osp.join(args.out_dir, 'bbox_smoothed', json_name)
data = gnu.load_json(json_path)
print(data)
break
else:
body_pose_list, body_bbox_list = body_bbox_detector.detect_body_pose(img_original_bgr)
break
hand_bbox_list = [None, ] * len(body_bbox_list)
if len(body_bbox_list) < 1:
print(f"No body deteced: {image_path}")
continue
#Sort the bbox using bbox size
# (to make the order as consistent as possible without tracking)
bbox_size = [ (x[2] * x[3]) for x in body_bbox_list]
idx_big2small = np.argsort(bbox_size)[::-1]
body_bbox_list = [ body_bbox_list[i] for i in idx_big2small ]
if args.single_person and len(body_bbox_list)>0:
body_bbox_list = [body_bbox_list[0], ]
# Body Pose Regression
pred_output_list = body_mocap.regress(img_original_bgr, body_bbox_list)
assert len(body_bbox_list) == len(pred_output_list)
# extract mesh for rendering (vertices in image space and faces) from pred_output_list
pred_mesh_list = demo_utils.extract_mesh_from_output(pred_output_list)
# visualization
res_img = visualizer.visualize(
img_original_bgr,
pred_mesh_list = pred_mesh_list,
body_bbox_list = body_bbox_list)
# show result in the screen
if not args.no_display:
res_img = res_img.astype(np.uint8)
ImShow(res_img)
# save result image
if args.out_dir is not None:
demo_utils.save_res_img(args.out_dir, image_path, res_img)
# save predictions to pkl
if args.save_pred_pkl:
demo_type = 'body'
demo_utils.save_pred_to_pkl(
args, demo_type, image_path, body_bbox_list, hand_bbox_list, pred_output_list)
timer.toc(bPrint=True,title="Time")
print(f"Processed : {image_path}")
#save images as a video
if not args.no_video_out and input_type in ['video', 'webcam']:
demo_utils.gen_video_out(args.out_dir, args.seq_name)
if input_type =='webcam' and input_data is not None:
input_data.release()
cv2.destroyAllWindows()