Object Discovery via Contrastive Learning for Weakly Supervised Object Detection

Weakly Supervised Object Detection (WSOD) is a task that detects objects in an image using model trained only on image-level annotations. Current state-of-the-art models benefit from self-supervised instance-level supervision, but since weak supervision does not include count or location information, the most common “argmax” labeling method often ignores many instances of objects. To alleviate this issue, we propose novel multiple instance called object discovery. We further introduce new contrastive loss under where no information available for sampling, weakly supervised (WSCL). WSCL aims to construct credible similarity threshold discovery by leveraging consistent features embedding vectors same class. As result, achieve results MS-COCO 2014 and 2017 as well PASCAL VOC 2012, competitive 2007. The code at https://github.com/jinhseo/OD-WSCL .