数据预处理阶段步骤4无法生成3dmm系数文件

[SoftSisterRui]

执行命令：python data_gen/utils/process_video/fit_3dmm_landmark.py --ds_name=nerf --vid_dir=data/raw/videos/khtest.mp4 --reset --debug --id_mode=global 看起来执行完成了但是没有生成3dmm系数文件 显卡:A40

[plzzzzg]

found the answer from another closed issue , just remove --debug when running fit_3dmm_landmark.py

when I exec cmd 'python data_gen/utils/process_video/fit_3dmm_landmark.py --ds_name=nerf --vid_dir=data/raw/videos/${VIDEO_ID}.mp4 --reset --debug --id_mode=global', the following message print:

reate rasterizer on device cuda:0cessed/videos/Lieu/debug_fit_3dmm.mp4..:   0%|                                                   | 0/250 [00:00<?, ?it/s]
rendering debug video to data/processed/videos/Lieu/debug_fit_3dmm.mp4..:   0%|                                                   | 0/250 [00:00<?, ?it/s]
data/raw/videos/Lieu.mp4 Not compiled with GPU support
finished 1 / 1 = 1.0000, failed 1 / 1 = 1.0000
process 0: fitting 3dmm ...: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:50<00:00, 50.85s/it]
all failed image names: ['data/raw/videos/Lieu.mp4']
All finished!

cr the code, the message only deal in debug mode.

    if debug:
        import imageio
        from utils.visualization.vis_cam3d.camera_pose_visualizer import CameraPoseVisualizer
        from data_util.face3d_helper import Face3DHelper
        from data_gen.utils.process_video.extract_blink import get_eye_area_percent
        face3d_helper = Face3DHelper('deep_3drecon/BFM', keypoint_mode='mediapipe')
    ...
        for i in tqdm.trange(min(250, len(frames)), desc=f'rendering debug video to {debug_name}..'):
            xy_cam3d_img = xy_camera_visualizer.extrinsic2pyramid(extrinsic[i], focal_len_scaled=0.25)

so exec the cmd without --debug solve my problem:

python data_gen/utils/process_video/fit_3dmm_landmark.py --ds_name=nerf --vid_dir=data/raw/videos/${VIDEO_ID}.mp4 --reset  --id_mode=global