Better detection of multiple objects when only using image-level labels with WSCL
Better detection of multiple objects when only using image-level labels with WSCL
Object Discovery via Contrastive Learning for Weakly Supervised Object Detection
arXiv paper abstract https://arxiv.org/abs/2208.07576v1
arXiv PDF paper https://arxiv.org/pdf/2208.07576v1.pdf
GitHub https://github.com/jinhseo/od-wscl
Weakly Supervised Object Detection (WSOD) is a task that detects objects in an image using a model trained only on image-level annotations.
… since weak supervision does not include count or location information, the most common ``argmax’’ labeling method often ignores many instances of objects.
… propose a novel multiple instance labeling method called object discovery.
… introduce a new contrastive loss under weak supervision where no instance-level information is available for sampling, called weakly supervised contrastive loss (WSCL).
WSCL aims to construct a credible similarity threshold for object discovery by leveraging consistent features for embedding vectors in the same class.
… achieve new state-of-the-art results on MS-COCO 2014 and 2017 as well as PASCAL VOC 2012, and competitive results on PASCAL VOC 2007.
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
