English

Few-Shot Segmentation via Cycle-Consistent Transformer

Computer Vision and Pattern Recognition 2022-03-09 v4

Abstract

Few-shot segmentation aims to train a segmentation model that can fast adapt to novel classes with few exemplars. The conventional training paradigm is to learn to make predictions on query images conditioned on the features from support images. Previous methods only utilized the semantic-level prototypes of support images as conditional information. These methods cannot utilize all pixel-wise support information for the query predictions, which is however critical for the segmentation task. In this paper, we focus on utilizing pixel-wise relationships between support and query images to facilitate the few-shot segmentation task. We design a novel Cycle-Consistent TRansformer (CyCTR) module to aggregate pixel-wise support features into query ones. CyCTR performs cross-attention between features from different images, i.e. support and query images. We observe that there may exist unexpected irrelevant pixel-level support features. Directly performing cross-attention may aggregate these features from support to query and bias the query features. Thus, we propose using a novel cycle-consistent attention mechanism to filter out possible harmful support features and encourage query features to attend to the most informative pixels from support images. Experiments on all few-shot segmentation benchmarks demonstrate that our proposed CyCTR leads to remarkable improvement compared to previous state-of-the-art methods. Specifically, on Pascal-5i5^i and COCO-20i20^i datasets, we achieve 67.5% and 45.6% mIoU for 5-shot segmentation, outperforming previous state-of-the-art methods by 5.6% and 7.1% respectively.

Keywords

Cite

@article{arxiv.2106.02320,
  title  = {Few-Shot Segmentation via Cycle-Consistent Transformer},
  author = {Gengwei Zhang and Guoliang Kang and Yi Yang and Yunchao Wei},
  journal= {arXiv preprint arXiv:2106.02320},
  year   = {2022}
}

Comments

Advances in Neural Information Processing Systems (NeurIPS), 2021. Project: https://github.com/GengDavid/CyCTR

R2 v1 2026-06-24T02:49:46.057Z