- 
          
- 
                Notifications
    You must be signed in to change notification settings 
- Fork 1.8k
Description
Is there an existing issue for this?
- I have searched the existing issues
Bug description
Hi, I've been analyzing my videos with
deeplabcut.analyze_videos(path_config_file, video, videotype='.avi', save_as_csv=True)
Most of the time, it's running well and outputs are all good.
But sometimes the following error pops out:
/home/adam/anaconda3/envs/DLC/lib/python3.9/site-packages/deeplabcut/pose_estimation_tensorflow/predict_videos.py:888: UserWarning: Could not decode frame #112637. warnings.warn(f"Could not decode frame #{counter}.") /home/adam/anaconda3/envs/DLC/lib/python3.9/site-packages/deeplabcut/pose_estimation_tensorflow/predict_videos.py:888: UserWarning: Could not decode frame #112638. warnings.warn(f"Could not decode frame #{counter}.") /home/adam/anaconda3/envs/DLC/lib/python3.9/site-packages/deeplabcut/pose_estimation_tensorflow/predict_videos.py:888: UserWarning: Could not decode frame #112639. warnings.warn(f"Could not decode frame #{counter}.")
It keeps going on until the last frame of the video. When I check the output and plotting, the frames starting from first line of this message all had likelihood value=0 for all labels, and all the labels have coordinates = 0.
This can be fixed by deleting the outputs and re-running the same line of code (sometimes keeps having the same issue but at different and random frames...).
So overall, it can be fixed but it's very annoying to detect which videos had the issue and re-run it when doing batch processing.
Operating System
Ubuntu 22.04.3 LTS
DeepLabCut version
dlc version 2.3.8
DeepLabCut mode
single animal
Device type
gpu NVIDIA GEFORCE RTX2080
Steps To Reproduce
No response
Relevant log output
No response
Anything else?
No response
Code of Conduct
- I agree to follow this project's Code of Conduct