Comment by stephanst

generalisation is 90% of the problem yep. This model is trained with a lot of augmentations and quite diverse data, but it still really needs fine tuning to specific use cases to get great efficiency unless you don’t care about false positives.