Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Error Analysis backend support for Object Detection #2225

Merged
merged 125 commits into from
Aug 28, 2023

Conversation

Advitya17
Copy link
Collaborator

@Advitya17 Advitya17 commented Aug 9, 2023

This PR adds EA support for the OD scenario, to achieve feature parity with the rest of the image dashboards. This baseline support with the error rate metric is equivalent in UX to that of the multilabel scenario.

This is also the first PR that utilizes the Error Labeling work from vision_explanation_methods repo to evaluate the prediction and consider OD-specific edge cases (such as localization, duplicate/missing detections etc). Further testing & improvements are separately being made to the Error Labeling work which EA can piggyback off of (and other parts of the dashboard in upcoming work).

Limitations:

  1. The IoU threshold is not currently surfaced to the user. Support could be added such as an IoU slider for EA, or an universal IoU slider after other components also utilize the Error Labeling work.
  2. The Error Rate metric is a naive metric for both OD & multilabel as it marks a prediction as incorrect even with 1 incorrect label/detection. Support for better metrics can be added based on customer feedback.

Snapshots after also making the change in #2233 + utilizing editable installs of responsibleai as per main:

image

image

image

Description

Checklist

  • I have added screenshots above for all UI changes.
  • I have added e2e tests for all UI changes.
  • Documentation was updated if it was needed.

@Advitya17 Advitya17 closed this Aug 9, 2023
@Advitya17 Advitya17 reopened this Aug 9, 2023
@Advitya17 Advitya17 marked this pull request as draft August 9, 2023 17:24
@natalie-isak

This comment was marked as resolved.

1 similar comment
@Advitya17 Advitya17 marked this pull request as ready for review August 28, 2023 14:44
@codecov-commenter
Copy link

codecov-commenter commented Aug 28, 2023

Codecov Report

Merging #2225 (532c180) into main (f22769f) will increase coverage by 5.51%.
Report is 1 commits behind head on main.
The diff coverage is n/a.

@@            Coverage Diff             @@
##             main    #2225      +/-   ##
==========================================
+ Coverage   88.94%   94.45%   +5.51%     
==========================================
  Files         133       18     -115     
  Lines        7532      415    -7117     
==========================================
- Hits         6699      392    -6307     
+ Misses        833       23     -810     
Flag Coverage Δ
unittests 94.45% <ø> (+5.51%) ⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

see 115 files with indirect coverage changes

3 similar comments
@Advitya17 Advitya17 enabled auto-merge (squash) August 28, 2023 17:27
@Advitya17 Advitya17 merged commit 21c7161 into main Aug 28, 2023
@Advitya17 Advitya17 deleted the agemawat/ea_od_backend branch August 28, 2023 17:52
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants