May I know how can I compute AUC value and ROC curve from an AutoML Vision trained model? Google does only provide the average precision and Precision-Recall curve for evaluation (including a confusion matrix). May I know are there methods in dealing this? Thanks so much!
At this moment AutoML Vision does not provide this feature to get AUC value and ROC, I suggest you to file a feature request asking for this feature.
Keep in mind that there is no ETA for approve/implement Feature Requests as is mentioned in the documentation.
After doing it I suggest you to “starring" that feature request to vote and supply feedback about how it would be useful. I also suggest you verify constantly the Cloud Vision release notes to know about all features, bug fixes, known issues, and deprecated functionality on AutoML Vision.