MaskSplit: Self-supervised Meta-learning for Few-shot Semantic Segmentation

TL;DR

This paper introduces MaskSplit, a self-supervised meta-learning approach for few-shot semantic segmentation that leverages unsupervised saliency estimation to generate pseudo-masks, reducing the need for manual annotations in training.

Contribution

It presents the first unsupervised method for few-shot segmentation on natural images using pseudo-masks and meta-learning, advancing the field of annotation-efficient segmentation.

Findings

Achieves promising results on few-shot segmentation tasks.

Demonstrates effectiveness of self-supervised training with pseudo-masks.

First to address unsupervised few-shot segmentation on natural images.

Abstract

Just like other few-shot learning problems, few-shot segmentation aims to minimize the need for manual annotation, which is particularly costly in segmentation tasks. Even though the few-shot setting reduces this cost for novel test classes, there is still a need to annotate the training data. To alleviate this need, we propose a self-supervised training approach for learning few-shot segmentation models. We first use unsupervised saliency estimation to obtain pseudo-masks on images. We then train a simple prototype based model over different splits of pseudo masks and augmentations of images. Our extensive experiments show that the proposed approach achieves promising results, highlighting the potential of self-supervised training. To the best of our knowledge this is the first work that addresses unsupervised few-shot segmentation problem on natural images.