In addition to testing YOLO/RetinaNet on our SEM dataset, I think we should also try it on one of our BF datasets. The detection task might actually be a bit easier there because the axon density is lower (whereas the SEM dataset has a pretty high density). This could give us interesting insight on how these models perform.
In addition to testing YOLO/RetinaNet on our SEM dataset, I think we should also try it on one of our BF datasets. The detection task might actually be a bit easier there because the axon density is lower (whereas the SEM dataset has a pretty high density). This could give us interesting insight on how these models perform.