CAST: Contrastive Adaptation and Distillation for Semi-Supervised Instance Segmentation

TL;DR

CAST introduces a semi-supervised distillation framework that effectively compresses large vision models into smaller, accurate instance segmentation models by leveraging contrastive learning and unlabeled data.

Contribution

The paper proposes a novel semi-supervised knowledge distillation method with an instance-aware contrastive loss for improved instance segmentation.

Findings

Smaller student models outperform zero-shot teachers by +8.5 and +7.1 AP.

CAST surpasses adapted teachers by +3.4 and +1.5 AP.

Outperforms state-of-the-art semi-supervised distillation methods on Cityscapes and ADE20K.

Abstract

Instance segmentation demands costly per-pixel annotations and computationally expensive models. We introduce CAST, a semi-supervised knowledge distillation (SSKD) framework that compresses pre-trained vision foundation models (VFM) into compact experts using limited labeled and abundant unlabeled data. CAST unfolds in three stages: (1) domain adaptation of the VFM(s) via self-training with contrastive calibration, (2) knowledge transfer through a unified multi-objective loss, and (3) student refinement to mitigate residual pseudo-label bias. Central to CAST is an \emph{instance-aware pixel-wise contrastive loss} that fuses mask and class scores to extract informative negatives and enforce clear inter-instance margins. By maintaining this contrastive signal across both adaptation and distillation, we align teacher and student embeddings and fully leverage unlabeled images. On Cityscapes and ADE20K, our ~11x smaller student improves over its zero-shot VFM teacher(s) by +8.5 and +7.1 AP, surpasses adapted teacher(s) by +3.4 and +1.5 AP, and further outperforms state-of-the-art SSKD methods on both benchmarks.