Solving Missing-Annotation Object Detection with Background Recalibration Loss
ICASSP, pp. 1888-1892, 2020.
This paper focuses on a novel and challenging detection scenario: A majority of true objects/instances is unlabeled in the datasets, so these missing-labeled areas will be regarded as the background during training. Previous art on this problem has proposed to use soft sampling to re-weight the gradients of RoIs based on the overlaps wi...More
PPT (Upload PPT)