PixCon: Clean-Positive Contrastive Learning for Foundation-Model Semi-Supervised Segmentation
2026-07-03 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionArtificial IntelligenceMachine Learning
AI summaryⓘ
The authors address how to improve semi-supervised semantic segmentation by focusing on which pixel labels to trust during training. They introduce PixCon, a method that only uses correctly classified pixels as positive examples, ensuring no wrong labels contaminate the learning process. Unlike previous methods that rely on filtering uncertain labels, PixCon keeps a clean set of positive pixels without extra complexity or thresholds. Their analysis shows that avoiding label contamination improves learning stability and yields consistent accuracy gains on standard benchmarks. The main benefit comes from cleaner learning signals, especially when using strong foundation-model teachers.
semi-supervised semantic segmentationpseudo-labelspixel contrastive learningmemory bankInfoNCE lossfoundation modelsDINOv2consistency regularizationpseudo-label contaminationmean Intersection over Union (mIoU)
Authors
Ebenezer Tarubinga
Abstract
Semi-supervised semantic segmentation (SSSS) has long turned on one question, which pseudo-labels to trust, and answered it with ever more careful confidence filtering. Foundation backbones change the regime: with a DINOv2 teacher a strict threshold already retains a measured 98%-clean pseudo-label set, so the accuracy that remains lives not in the filter but in how the embedding space is structured by class. We propose PixCon, a clean-positive pixel-contrastive framework. PixCon maintains a per-class memory bank that admits only labeled pixels the student already classifies correctly, guaranteeing a contamination-free positive set ($ρ_F=0$) by construction, unlike prior contrastive SSSS banks (ReCo, U$^2$PL) built from confidence-filtered pseudo-labels. It is a single branch over a consistency backbone, adds no inference-time parameters, and needs no bank-specific threshold. A first-order analysis of the supervised-InfoNCE gradient explains why contamination hurts: its false-positive term scales as $ρ_F/(1-ρ_F)$, which we measure (0.018 on Pascal, 0.106 on ADE20K) rather than assume. Across Pascal VOC, Cityscapes, and ADE20K, PixCon matches or improves a strong DINOv2-based UniMatch V2 baseline in a compute-matched one-switch protocol: it improves every Pascal-1/8 seed (a per-seed gain of about +0.2 mIoU) and its three-seed mean reaches 87.90, the published UniMatch V2-B figure. Because contamination is already rare under foundation-model teachers, our analysis indicates the $ρ_F=0$ guarantee acts chiefly as robustness as teachers weaken, while the accuracy gain comes from cleaner positive supervision, making clean-positive contrast a robust, low-cost default for foundation-model SSSS.