Exif Nested Metadata support for Image Dashboards

[Advitya17]

This pull request includes changes to the responsibleai_vision package, specifically to the feature_extractors.py and image_reader.py files. The changes aim to improve the extraction and processing of image metadata. The most important changes include the addition of ExifTags and IFDRational from the PIL library, the creation of a new function process_data to handle different data types, and the modification of the append_exif_features and get_all_exif_feature_names functions to handle nested tags in the Exif data.

Improvements to image metadata extraction and processing:

• responsibleai_vision/responsibleai_vision/utils/feature_extractors.py: Imported ExifTags and IFDRational from the PIL library to handle additional metadata types. Created a new function process_data to handle different data types in the Exif data. Modified the append_exif_features function to handle nested tags in the Exif data. [1] [2]

• responsibleai_vision/responsibleai_vision/utils/image_reader.py: Imported ExifTags from the PIL library and created a dictionary IFD_CODE_LOOKUP to map IFD block tags. Modified the get_all_exif_feature_names function to handle nested tags in the Exif data. [1] [2] [3]

Description

Ask from 2 customers to expand exif metadata support by including nested types.

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.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 90.32258% with 3 lines in your changes are missing coverage. Please review.

Project coverage is 78.15%. Comparing base (7292ba3) to head (f8a2902).

Files	Patch %	Lines
...n/responsibleai_vision/utils/feature_extractors.py	86.36%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2542      +/-   ##
==========================================
- Coverage   83.96%   78.15%   -5.81%     
==========================================
  Files          95       36      -59     
  Lines        4951     2619    -2332     
==========================================
- Hits         4157     2047    -2110     
+ Misses        794      572     -222     

Flag	Coverage Δ
unittests	78.15% <90.32%> (-5.81%)	⬇️

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

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[imatiach-msft]

I think these come from a settings file for each dataset (singleFeatureCohorts or multiFeatureCohorts). I think it might be better to just set these to the correct values for the vision notebooks or only do this type of very loose check for vision notebooks only, since we still want to validate this is consistent for tabular and text notebooks.

[Advitya17]

@imatiach-msft these tests are unfortunately so flaky that different python versions were showing different values and hence failing at different steps. For diff versions, this gate (Exif Nested Metadata support for Image Dashboards · microsoft/responsible-ai-toolbox@34f2f52 (github.com) has failed differently across all 3 notebooks.