You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Should the annotator really be decoupled from the analyzer? This would mean to store a lot information about single bugs like species, REAL coordinates (regarding that they are probably found on a cropped and scaled image) and image URI.
The pipeline should take an image and a annotation-helper as input and provide a method to select which algorithm(s) should be executed on that image.
The text was updated successfully, but these errors were encountered: