cannot pickle 'memoryview' object on pytorch deeplabcut #2739
Description
Is there an existing issue for this?
- I have searched the existing issues
Bug description
After training a model using the deeplabcut GUI and then running Analyze videos on any video, I always get this error:
(DEEPLABCUT) jake@dep65287:~/projects/hyperiid_dlc$ python -m deeplabcut
Loading DLC 3.0.0rc4...
Starting GUI...
QPixmap::scaled: Pixmap is a null pixmap
INFO:console:Analyzing videos with /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/dlc-models-pytorch/iteration-0/HyperiidMay31-trainset95shuffle1/train/snapshot-200.pt
INFO:console: -> Using detector /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/dlc-models-pytorch/iteration-0/HyperiidMay31-trainset95shuffle1/train/snapshot-detector-200.pt
INFO:console:Starting to analyze /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippet.mp4
INFO:console:Video metadata:
Overall # of frames: 930
Duration of video [s]: 31.04
fps: 29.96
resolution: w=950, h=980
INFO:console:Running detector with batch size 1
100%|█████████████████████████████████████████████████████████████████████████▉| 929/930 [00:23<00:00, 39.60it/s]
0%| | 0/930 [00:00<?, ?it/s]INFO:console:Running pose prediction with batch size 32
100%|█████████████████████████████████████████████████████████████████████████▉| 929/930 [00:39<00:00, 23.66it/s]
WARNING:root:The video metadata indicates that there 930 in the video, but only 929 were able to be processed. This can happen if the video is corrupted. You can try to fix the issue by re-encoding your video (tips on how to do that: https://deeplabcut.github.io/DeepLabCut/docs/recipes/io.html#tips-on-video-re-encoding-and-preprocessing)
INFO:console:Saving results in /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200.h5 and /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200_full.pickle
INFO:console:Processing...
INFO:console:
INFO:console:/home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippet.mp4
INFO:console:Loading From /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200.h5
929it [00:00, 1004772.67it/s]
Traceback (most recent call last):
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/gui/utils.py", line 22, in run
self.func()
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/compat.py", line 865, in analyze_videos
return analyze_videos(
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/pose_estimation_pytorch/apis/analyze_videos.py", line 384, in analyze_videos
convert_detections2tracklets(
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/pose_estimation_pytorch/apis/convert_detections_to_tracklets.py", line 241, in convert_detections2tracklets
pickle.dump(tracklets, f, pickle.HIGHEST_PROTOCOL)
TypeError: cannot pickle 'memoryview' object
I don't get this error on the older tensorflow version of the software. However, I have extreme trouble getting that to work with my GPU, so that's why I updated to the pytorch 3.0.0 version.
Operating System
Ubuntu 20.04.6 LTS
DeepLabCut version
3.0.0rc4
DeepLabCut mode
multi animal
Device type
NVIDIA GeForce RTX 3090
Steps To Reproduce
- Using the conda DEEPLABCUT.yaml from the instructions, install to conda
- Activate the environment.
- Train a model using pytorch environment
- Run analyze videos using any number of videos
Relevant log output
(DEEPLABCUT) jake@dep65287:~/projects/hyperiid_dlc$ python -m deeplabcut
Loading DLC 3.0.0rc4...
Starting GUI...
QPixmap::scaled: Pixmap is a null pixmap
INFO:console:Analyzing videos with /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/dlc-models-pytorch/iteration-0/HyperiidMay31-trainset95shuffle1/train/snapshot-200.pt
INFO:console: -> Using detector /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/dlc-models-pytorch/iteration-0/HyperiidMay31-trainset95shuffle1/train/snapshot-detector-200.pt
INFO:console:Starting to analyze /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippet.mp4
INFO:console:Video metadata:
Overall # of frames: 930
Duration of video [s]: 31.04
fps: 29.96
resolution: w=950, h=980
INFO:console:Running detector with batch size 1
100%|█████████████████████████████████████████████████████████████████████████▉| 929/930 [00:23<00:00, 39.60it/s]
0%| | 0/930 [00:00<?, ?it/s]INFO:console:Running pose prediction with batch size 32
100%|█████████████████████████████████████████████████████████████████████████▉| 929/930 [00:39<00:00, 23.66it/s]
WARNING:root:The video metadata indicates that there 930 in the video, but only 929 were able to be processed. This can happen if the video is corrupted. You can try to fix the issue by re-encoding your video (tips on how to do that: https://deeplabcut.github.io/DeepLabCut/docs/recipes/io.html#tips-on-video-re-encoding-and-preprocessing)
INFO:console:Saving results in /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200.h5 and /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200_full.pickle
INFO:console:Processing...
INFO:console:
INFO:console:/home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippet.mp4
INFO:console:Loading From /home/jake/projects/hyperiid_dlc/dlc_projects/hyperia-Jake-2023-05-31/videos/16L-processed-snippetDLC_AnimaltokenposeBase_HyperiidMay31shuffle1_detector_200_snapshot_200.h5
929it [00:00, 1004772.67it/s]
Traceback (most recent call last):
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/gui/utils.py", line 22, in run
self.func()
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/compat.py", line 865, in analyze_videos
return analyze_videos(
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/pose_estimation_pytorch/apis/analyze_videos.py", line 384, in analyze_videos
convert_detections2tracklets(
File "/home/jake/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/deeplabcut/pose_estimation_pytorch/apis/convert_detections_to_tracklets.py", line 241, in convert_detections2tracklets
pickle.dump(tracklets, f, pickle.HIGHEST_PROTOCOL)
TypeError: cannot pickle 'memoryview' objectAnything else?
No response
Code of Conduct
- I agree to follow this project's Code of Conduct