Improving Dense Contrastive Learning with Dense Negative Pairs
This work addresses the need for enhanced representation learning in computer vision tasks requiring precise spatial localization, but it is incremental as it builds directly on existing methods like DenseCL.
The paper tackles the problem of improving dense contrastive learning for better spatial feature localization in tasks like multi-label classification and segmentation, resulting in a 3.5% mAP improvement over SimCLR and 4% over DenseCL in COCO multi-label classification, and 1.8% and 0.7% mIoU gains in COCO and VOC segmentation.
Many contrastive representation learning methods learn a single global representation of an entire image. However, dense contrastive representation learning methods such as DenseCL (Wang et al., 2021) can learn better representations for tasks requiring stronger spatial localization of features, such as multi-label classification, detection, and segmentation. In this work, we study how to improve the quality of the representations learned by DenseCL by modifying the training scheme and objective function, and propose DenseCL++. We also conduct several ablation studies to better understand the effects of: (i) various techniques to form dense negative pairs among augmentations of different images, (ii) cross-view dense negative and positive pairs, and (iii) an auxiliary reconstruction task. Our results show 3.5% and 4% mAP improvement over SimCLR (Chen et al., 2020a) andDenseCL in COCO multi-label classification. In COCO and VOC segmentation tasks, we achieve 1.8% and 0.7% mIoU improvements over SimCLR, respectively.