Uncertainty-Aware Semi-Supervised Few Shot Segmentation

TL;DR

This paper introduces a semi-supervised approach for few shot segmentation that utilizes uncertainty-guided pseudo labels from unlabeled images to improve segmentation accuracy, demonstrating significant performance gains on standard benchmarks.

Contribution

It proposes a novel semi-supervised FSS method that jointly predicts segmentation and estimates uncertainty, enabling reliable pseudo label generation without extra training.

Findings

Effective removal of unreliable pseudo labels improves segmentation quality.

Significant performance improvements over state-of-the-art on PASCAL-5^i and COCO-20^i.

Method is end-to-end and compatible with existing approaches.

Abstract

Few shot segmentation (FSS) aims to learn pixel-level classification of a target object in a query image using only a few annotated support samples. This is challenging as it requires modeling appearance variations of target objects and the diverse visual cues between query and support images with limited information. To address this problem, we propose a semi-supervised FSS strategy that leverages additional prototypes from unlabeled images with uncertainty guided pseudo label refinement. To obtain reliable prototypes from unlabeled images, we meta-train a neural network to jointly predict segmentation and estimate the uncertainty of predictions. We employ the uncertainty estimates to exclude predictions with high degrees of uncertainty for pseudo label construction to obtain additional prototypes based on the refined pseudo labels. During inference, query segmentation is predicted using prototypes from both support and unlabeled images including low-level features of the query images. Our approach is end-to-end and can easily supplement existing approaches without the requirement of additional training to employ unlabeled samples. Extensive experiments on PASCAL-$5^i$ and COCO-$20^i$ demonstrate that our model can effectively remove unreliable predictions to refine pseudo labels and significantly improve upon state-of-the-art performances.