search

facebookresearch / VideoPose3D

Efficient 3D human pose estimation in video using 2D keypoint trajectories
Other
3.71k stars 753 forks source link

run.py can not run training from scratch command #179

Open JinluZhang1126 opened 3 years ago

JinluZhang1126 commented 3 years ago

Good afternoon,

 I have some problems when I tried to run `python run.py -e 80 -k cpn_ft_h36m_dbb -arc 3,3,3,3,3` to test the training function of VideoPose3D. The  problem occurred is list below:

I have set up as README.md, and everything is well.

And then I run the command in the README.md( `python run.py -e 80 -k cpn_ft_h36m_dbb -arc 3,3,3,3,3` ) to train the model. It could load the model and load the dataset. But it stuck in a step, I tried several times, and it always stuck at the same step for a long time(I use Ctrl+C to stop the program after a few hours), I wish anybody could do me a favour.👀

Namespace(actions='*', architecture='3,3,3,3,3', batch_size=1024, bone_length_term=True, by_subject=False, causal=False, channels=1024, checkpoint='checkpoint', checkpoint_frequency=10, data_augmentation=True, dataset='h36m', dense=False, disable_optimizations=False, downsample=1, dropout=0.25, epochs=80, evaluate='', export_training_curves=False, keypoints='cpn_ft_h36m_dbb', learning_rate=0.001, linear_projection=False, lr_decay=0.95, no_eval=False, no_proj=False, render=False, resume='', stride=1, subjects_test='S9,S11', subjects_train='S1,S5,S6,S7,S8', subjects_unlabeled='', subset=1, test_time_augmentation=True, viz_action=None, viz_bitrate=3000, viz_camera=0, viz_downsample=1, viz_export=None, viz_limit=-1, viz_no_ground_truth=False, viz_output=None, viz_size=5, viz_skip=0, viz_subject=None, viz_video=None, warmup=1) Loading dataset... Preparing data... Loading 2D detections... INFO: Receptive field: 243 frames INFO: Trainable parameter count: 16952371 INFO: Testing on 543344 frames INFO: Training on 1559752 frames Note: reported losses are averaged over all frames and test-time augmentation is not used here. The final evaluation will be carried out after the last training epoch. ^CTraceback (most recent call last): File "run.py", line 405, in inputs_2d = torch.from_numpy(batch_2d.astype('float32')) KeyboardInterrupt

jionie commented 3 years ago

You can set --export-training-curves to save training logs in picture, and seems like while training it will not output training loss, and it takes time to finish one epoch. And you can also use nvidia-smi to see if GPU is using.