IssueTracker

This feature request is to include the possibility of modifying Vision API logo detection [1] to include custom set of new logos not included in GCP default's model (using a custom training set for the new logos).

That is, if Vision API logo detection detects x number of logos and newLogoA is not detected as is not present in this set of x logos, it should be possible to make a custom model on top of vision API logo detection to detect x logos plus the newLogoA.

It is currently possible to train a custom model from scratch, but this will not include all the logos already recognize by the default vision API logo detection model.

[1] https://cloud.google.com/vision/docs/detecting-logos