search

tnc-ca-geo / animl-frontend

A frontend web app for viewing & labeling camera trap data by The Nature Conservancy.
https://animl.camera
Other
16 stars 0 forks source link

Make creation & invalidation of "empty" labels more inteligent / automatic #77

Open nathanielrindlaub opened 2 years ago

nathanielrindlaub commented 2 years ago

Right now, an image can have no objects at all for two reasons: it either never received a prediction from Megadetector, or it did get predictions, but the returned objects/labels we all invalidated. If all objects on an image a manually invalidated, the user still has to click "mark as empty" to apply an empty object to that image. It would be nice to automate this.

Likewise, if Megadetector didn't find any objects and an empty object was applied, and then a user adds an object to it, it would be great to automatically invalidate the empty object.

nathanielrindlaub commented 1 year ago

The question is: what are the preconditions for running this "are there objects in the image" check and auto-applying or removing empty labels? It could happen

  1. after every object removal/addition
  2. after an image gets blurred

Also still need to be mindful/think about how this would interact w/ undo & redo.

Also need to think about what controls would be necessary after implementing this (i.e., is "Mark empty" still necessary).