Pose Extraction Error

[zhu-ch]

System Info / 系統信息

Python 3.10

Who can help? / 谁可以帮助到您？

No response

Information / 问题信息

• The official example scripts / 官方的示例脚本
• My own modified scripts / 我自己修改的脚本和任务

Reproduction / 复现过程

Thanks a lot for your work! But I found I error when extract pose with process_pose.py

Processing driving video...
Detecting poses in video: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 144/144 [00:08<00:00, 16.22it/s]
Running NLF on driving video...
Processing reference image...
Running NLF on reference image...
Aligning and rendering...
Traceback (most recent call last):
  File "/home/SCAIL/scail_pose/NLFPoseExtract/process_pose.py", line 246, in <module>
    pose_3d_first_driving_frame = nlf_results[0]['nlfpose'][0][0].cpu().numpy()  # 3D点 frame-idx bbox-idx detect-idx
IndexError: index 0 is out of bounds for dimension 0 with size 0

The ref_img and video are:
https://huggingface.co/datasets/theFoxofSky/RealisDance-Val/resolve/main/ref/0027.png
https://huggingface.co/datasets/theFoxofSky/RealisDance-Val/resolve/main/gt/0027.mp4

Could you please help me to fix it? Thanks a lot!

Expected behavior / 期待表现

fix it

[teal024]

Seems to be an issue when nlf failed to detect the the first frame of the driving sequence and retarget is enabled.

[zhu-ch]

But there are clearly visible people in the video.
And, process_pose_multi.py could run this case

[teal024]

But there are clearly visible people in the video. And, process_pose_multi.py could run this case

Sry, it's more of an issue when retarget is enabled and nlf failed to detect the the first frame, not the whole sequence. We'll fix that, thank you for your feedback!!