Bootstrapping Semantic Segmentation with Regional Contrast

Shikun Liu · Shuaifeng Zhi · Edward Johns · Andrew Davison


Keywords: [ semi-supervised learning ] [ contrastive learning ] [ semantic segmentation ]

[ Abstract ]
[ Visit Poster at Spot F0 in Virtual World ] [ Slides [ OpenReview
Mon 25 Apr 2:30 a.m. PDT — 4:30 a.m. PDT


We present ReCo, a contrastive learning framework designed at a regional level to assist learning in semantic segmentation. ReCo performs pixel-level contrastive learning on a sparse set of hard negative pixels, with minimal additional memory footprint. ReCo is easy to implement, being built on top of off-the-shelf segmentation networks, and consistently improves performance, achieving more accurate segmentation boundaries and faster convergence. The strongest effect is in semi-supervised learning with very few labels. With ReCo, we achieve high quality semantic segmentation model, requiring only 5 examples of each semantic class.

Chat is not available.