Prototype Mixture Models for Few-shot Semantic Segmentation

TL;DR

This paper introduces prototype mixture models (PMMs) for few-shot semantic segmentation, leveraging multiple prototypes and an EM algorithm to better capture diverse object appearances and improve segmentation accuracy.

Contribution

The paper proposes PMMs that use multiple prototypes and EM optimization to enhance semantic representation in few-shot segmentation, outperforming existing methods.

Findings

PMMs significantly outperform state-of-the-art on Pascal VOC and MS-COCO.

PMMs improve 5-shot segmentation on MS-COCO by up to 5.82%.

PMMs achieve a good balance between accuracy and computational cost.

Abstract

Few-shot segmentation is challenging because objects within the support and query images could significantly differ in appearance and pose. Using a single prototype acquired directly from the support image to segment the query image causes semantic ambiguity. In this paper, we propose prototype mixture models (PMMs), which correlate diverse image regions with multiple prototypes to enforce the prototype-based semantic representation. Estimated by an Expectation-Maximization algorithm, PMMs incorporate rich channel-wised and spatial semantics from limited support images. Utilized as representations as well as classifiers, PMMs fully leverage the semantics to activate objects in the query image while depressing background regions in a duplex manner. Extensive experiments on Pascal VOC and MS-COCO datasets show that PMMs significantly improve upon state-of-the-arts. Particularly, PMMs improve 5-shot segmentation performance on MS-COCO by up to 5.82\% with only a moderate cost for model size and inference speed.