Advanced to next tabs after running pose segmentation

[katiekly]

In the GUI, I was able to successfully run pose segmentation but I wasn't able to advance to the next steps such as Motif Videos tab and UMAP Visualization tab.

[luiztauffer]

@katiekly could you show me exactly where you're getting stuck? The tabs do not show the results? Or the app crashes?
Can you please also check if the results files were correctly created in their folders?
Which OS you're using?

[katiekly]

The tabs isn't supposed to show the results, but I'm not able proceed to the next steps after pose segmentation, so clicking on the Motif Videos or UMAP Visualization tab doesn't do anything because the app still thinks I have not completed pose segmentation even though I have. I checked the results folder and all expected outputs are there, but there's a hmm_trained.pkl file in the immediate results folder that shouldn't be there. This was tested on Windows 11 Pro Education

[luiztauffer]

Could you share the log files?

[katiekly]

create_trainset.log
egocentric_alignment.log
evaluate_model.log
pose_segmentation.log
train_model.log

[DrSRMiller]

@luiztauffer We also notice that "states.json" is showing segmentation as "running" even though it has successfully completed. We tried to manually advance the project state by editing the "states.json" file but we were not able to advance or backtrack states. We were not able to manually modify project.workflow because we couldn't find it.

[luiztauffer]

@katiekly @DrSRMiller are you be able to run the pipeline without the GUI, just using the package? That can help us isolate the problem

[luiztauffer]

@nathan-vm

[DrSRMiller]

No, it is a demo of the GUI specifically.

…

On Thu, Aug 15, 2024 at 11:54 PM Luiz Tauffer ***@***.***> wrote: @nathan-vm <https://github.com/nathan-vm> — Reply to this email directly, view it on GitHub <#49 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A3JD67S26GVO4WDCE62SY3LZRWOZ7AVCNFSM6AAAAABMRFHBX2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEOJSHEZTCMJYGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[luiztauffer]

@DrSRMiller I mean, are you able to successfully the pipeline with this specific dataset, using only the package? I want to isolate the problem, if it runs fine with just the package, it might be a bug on the desktop app. If not, it might be an edge case we haven't encountered before on vame processing itself.
Could you shared this data so we can try to reproduce the error?

[katiekly]

@luiztauffer I shared the project folder where I encountered the error with you via email. Please let me know if you didn't receive the invite to the drive.

[luiztauffer]

@katiekly can you please also share with:

• nathan.v.marcelino@gmail.com
• vinicius@taufferconsulting.com

[bendichter]

I thought I was encountering the same error. It was stuck on that page for a while. The log seemed to indicate that the pipeline was complete, and the pose segmentation button was inactive, but I could not progress to the next page.

The log at that point was ending in:

2024-08-16 14:58:30.212 INFO --- [Thread-856 (process_request_thread)] vame.analysis.pose_segmentation : 279 : For all animals the same parametrization of latent vectors is applied for 15 cluster

I thought this was the end, but it was not. After several minutes it changed to add:

2024-08-16 14:58:30.213 INFO --- [Thread-856 (process_request_thread)] vame.analysis.pose_segmentation : 135 : Using kmeans as parametrization!
2024-08-16 14:58:36.336 INFO --- [Thread-856 (process_request_thread)] vame.analysis.pose_segmentation : 317 : /Users/bendichter/vame-desktop/projects/test-Aug16-2024/results/video-1/VAME/kmeans-15/
2024-08-16 14:58:36.337 INFO --- [Thread-856 (process_request_thread)] vame.analysis.pose_segmentation : 332 : You succesfully extracted motifs with VAME! From here, you can proceed running vame.motif_videos()

and the app screen changed to:

and allowed me to progress. @katiekly could you try leaving the app for ~15 minutes to see if you are in the same situation? If so, I think we just need a minor UI modification to make it clear that the app is not hanging.