CA-SSL: Class-Agnostic Semi-Supervised Learning for Detection and Segmentation. (arXiv:2112.04966v2 [cs.CV] UPDATED)

To improve instance-level detection/segmentation performance, existing
self-supervised and semi-supervised methods extract either task-unrelated or
task-specific training signals from unlabeled data. We show that these two
approaches, at the two extreme ends of the task-specificity spectrum, are
suboptimal for the task performance. Utilizing too little task-specific
training signals causes underfitting to the ground-truth labels of downstream
tasks, while the opposite causes overfitting to the ground-truth labels. To
this end, we propose a novel Class-Agnostic Semi-Supervised Learning (CA-SSL)
framework to achieve a …

arxiv cv detection learning segmentation semi-supervised semi-supervised learning ssl supervised learning