BiOpt: Bi-Level Optimization for Few-Shot Segmentation

TL;DR

This paper introduces BiOpt, a bi-level optimization approach for few-shot segmentation that leverages query image information to improve prototype learning, achieving state-of-the-art results on standard benchmarks.

Contribution

The paper proposes a novel bi-level optimization framework that computes class prototypes from query images in an inductive setting, enhancing few-shot segmentation performance.

Findings

Achieves state-of-the-art results on 5-shot PASCAL-5^i.

Achieves state-of-the-art results on 1-shot COCO-20^i.

Demonstrates the effectiveness of bi-level optimization in few-shot segmentation.

Abstract

Few-shot segmentation is a challenging task that aims to segment objects of new classes given scarce support images. In the inductive setting, existing prototype-based methods focus on extracting prototypes from the support images; however, they fail to utilize semantic information of the query images. In this paper, we propose Bi-level Optimization (BiOpt), which succeeds to compute class prototypes from the query images under inductive setting. The learning procedure of BiOpt is decomposed into two nested loops: inner and outer loop. On each task, the inner loop aims to learn optimized prototypes from the query images. An init step is conducted to fully exploit knowledge from both support and query features, so as to give reasonable initialized prototypes into the inner loop. The outer loop aims to learn a discriminative embedding space across different tasks. Extensive experiments on two benchmarks verify the superiority of our proposed BiOpt algorithm. In particular, we consistently achieve the state-of-the-art performance on 5-shot PASCAL-$5^i$ and 1-shot COCO-$20^i$.