talmolab / sleap

A deep learning framework for multi-animal pose tracking.
https://sleap.ai
Other
435 stars 97 forks source link

Add video path and frame indices to metrics #1396

Closed roomrys closed 1 year ago

roomrys commented 1 year ago

Description

Multiple people have asked if there might be a way to link back to frames where the model has performed poorly on as a way of either determining whether a certain video is culprit or mislabeled instances are culprit. This PR specifically follows the code provided by @adam-matic in #1253.

While this PR does not tie distances to the instances like a Dataframe might accomplish, it returns the instances (in a dictionary) in the same order as the distances. Feel free to suggest improvements!

Types of changes

Does this address any currently open issues?

Outside contributors checklist

Thank you for contributing to SLEAP!

:heart:

codecov[bot] commented 1 year ago

Codecov Report

Merging #1396 (fd084af) into develop (fb61b6c) will increase coverage by 0.17%. The diff coverage is 100.00%.

@@             Coverage Diff             @@
##           develop    #1396      +/-   ##
===========================================
+ Coverage    72.78%   72.95%   +0.17%     
===========================================
  Files          133      133              
  Lines        23718    23726       +8     
===========================================
+ Hits         17263    17310      +47     
+ Misses        6455     6416      -39     
Impacted Files Coverage Δ
sleap/nn/training.py 75.79% <ø> (ø)
sleap/nn/evals.py 92.46% <100.00%> (+17.46%) :arrow_up:

... and 1 file with indirect coverage changes

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more