fix image downloader failing with automl format on deserialize due to missing label transformations

[imatiach-msft]

Description

fix image downloader failing with automl format on deserialize due to missing label transformations

Seeing errors when deserializing RAIVisionInsights with automl format labels:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/azureml-envs/responsibleai-vision/lib/python3.8/site-packages/raiwidgets/responsibleai_dashboard.py", line 36, in __init__
    self.input = ResponsibleAIDashboardInput(
  File "/azureml-envs/responsibleai-vision/lib/python3.8/site-packages/raiwidgets/responsibleai_dashboard_input.py", line 41, in __init__
    self.dashboard_input = analysis.get_data()
  File "/azureml-envs/responsibleai-vision/lib/python3.8/site-packages/responsibleai_vision/rai_vision_insights/rai_vision_insights.py", line 558, in get_data
    dataset = self._get_dataset()
  File "/azureml-envs/responsibleai-vision/lib/python3.8/site-packages/responsibleai_vision/rai_vision_insights/rai_vision_insights.py", line 696, in _get_dataset
    dashboard_dataset.true_y = self._format_od_labels(
  File "/azureml-envs/responsibleai-vision/lib/python3.8/site-packages/responsibleai_vision/rai_vision_insights/rai_vision_insights.py", line 794, in _format_od_labels
    object_index = int(detection[0] - 1)
KeyError: 0

After extensive debugging, we discovered this was in the deserializer for true_y labels when image downloader is specified and the test dataset is reset, but the labels are not converted, which we could confirm based on the printed labels when adding debug:

...
detection:  [3.0, 296.3221740722656, 169.22532653808594, 383.11572265625, 305.3504333496094, 0.6043946146965027]
detection:  [2.0, 164.48257446289062, 220.76792907714844, 471.8297119140625, 508.36151123046875, 0.8620063662528992]
detection:  {'label': 'carton', 'topX': 0.2004008016032064, 'topY': 0.25975975975975973, 'bottomX': 0.46693386773547096, 'bottomY': 0.7822822822822822, 'isCrowd': 0}

The first two detections are in the converted common format for predicted labels. The third detection is for the true_y labels and is in the automl format and was not transformed. Transforming the labels on deserialize fixed the error.
Also new tests for object detection serialize + deserialize were added as part of this PR, which only previously existed for other tasks like image classification.

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.

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html

[github-actions]

https://responsibleai.blob.core.windows.net/pullrequest/microsoft/responsible-ai-toolbox/ilmat/fix-image-downloader-with-automl-format-deserialize/dashboard/index.html