1 paper · 1 filter
Kerri Lu, Dan M. Kluger, Stephen Bates +1
Current instance segmentation models achieve high performance on average predictions, but lack principled uncertainty quantification: their outputs are not calibrated, and there is…