2分半钟的视频，训练时候报错

[touwenameng]

Traceback (most recent call last):
File "train_mouth.py", line 328, in
training(lp.extract(args), op.extract(args), pp.extract(args), args.test_iterations, args.save_iterations, args.checkpoint_iterations, args.start_checkpoint, args.debug_from)
File "train_mouth.py", line 49, in training
scene = Scene(dataset, gaussians)
File "/root/autodl-tmp/TalkingGaussian/scene/init.py", line 47, in init
scene_info = sceneLoadTypeCallbacks["Blender"](args.source_path, args.white_background, args.eval, args=args)
File "/root/autodl-tmp/TalkingGaussian/scene/dataset_readers.py", line 256, in readNerfSyntheticInfo
test_cam_infos = readCamerasFromTransforms(path, "transforms_val.json", white_background, extension, audio_file, audio_extractor)
File "/root/autodl-tmp/TalkingGaussian/scene/dataset_readers.py", line 211, in readCamerasFromTransforms
talking_dict['blink'] = torch.as_tensor(np.clip(au_blink[frame['img_id']], 0, 2) / 2)
IndexError: index 3612 is out of bounds for axis 0 with size 3612

2分半钟的视频，按照流程准备完数据集，跑训练的时候直接报index数组错误，从报错上看像是验证集数据越界，请问这有可能啥原因呢

[Fictionarry]

没有遇到过这个问题，不太清楚具体是什么原因造成的，可能是帧率之类的问题，可以检查一下au.csv中的项数是否与帧数一致

[touwenameng]

好的，我再测试一下

[touwenameng]

帧率是对上了，但是process的时候又报错了，增加了GPU和内存都不行