How to evaluate on tiktok dataset?

Hi, its a good job and I am trying to reproduce your work but Sorry I didn't tried a lot of times but I cannot use this repo on tiktok dataset. How you evaluate your model on tiktok? How could I reproduce your metrics reported in your paper? BTW, what is the exactly 10 videos you choose in your evaluation? ``` (UniAnimate) root@miles:/mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate# python run_align_pose.py --ref_name /root/milesdata/BenchMark/TikTokdataset/00002/images/0001.png --source_video_paths /root/milesdata/BenchMark/TikTokdataset/00002/00002.mp4 --saved_pose_dir /root/milesdata/BenchMark/TikTokdataset/00002/save_pose/ 2025-04-08 15:33:31,330 - dw pose extraction - INFO - There are 1 videos for extracting poses 2025-04-08 15:33:31,330 - dw pose extraction - INFO - LOAD: DW Pose Model 2025-04-08 15:33:34.335279395 [W:onnxruntime:, session_state.cc:1166 VerifyEachNodeIsAssignedToAnEp] Some nodes were not assigned to the preferred execution providers which may or may not have an negative impact on performance. e.g. ORT explicitly assigns shape related ops to CPU to improve perf. 2025-04-08 15:33:34.335309706 [W:onnxruntime:, session_state.cc:1168 VerifyEachNodeIsAssignedToAnEp] Rerunning with verbose output on a non-minimal build will show node assignments. 2025-04-08 15:33:34,390 - dw pose extraction - INFO - 0/1, /root/milesdata/BenchMark/TikTokdataset/00002/00002.mp4 2025-04-08 15:33:49,348 - dw pose extraction - INFO - all frames in /root/milesdata/BenchMark/TikTokdataset/00002/00002.mp4 have been read. 223 /mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py:355: RuntimeWarning: invalid value encountered in scalar divide arm4_ratio = l_arm4_ref / l_arm4_0 /mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py:466: RuntimeWarning: invalid value encountered in scalar divide ll2_ratio = l_ll2_ref / l_ll2_0 /mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py:493: RuntimeWarning: invalid value encountered in scalar divide rl2_ratio = l_rl2_ref / l_rl2_0 Traceback (most recent call last): File "/mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py", line 712, in <module> mp_main(args) File "/mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py", line 689, in mp_main dwpose_woface, dwpose_wface = draw_pose(results_vis[i], H=768, W=512) File "/mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/run_align_pose.py", line 178, in draw_pose canvas = util.draw_handpose(canvas, hands) File "/mnt/jfs/milesdata/BenchMark/benchmethods/UniAnimate/dwpose/util.py", line 168, in draw_handpose x1 = int(x1 * W) ValueError: cannot convert float NaN to integer ``` This kind of problems also occur when I change 00010 to 00002/00003/00004/00005/00006 But I can successfully run the cmd in readme, it's not the due to environmental settings I think.