Improving Few-Shot Part Segmentation using Coarse Supervision

TL;DR

This paper introduces a framework that leverages coarse supervision signals like figure-ground masks and keypoints to enhance part segmentation models, reducing annotation costs and handling diverse labeling styles.

Contribution

It proposes a joint learning approach to connect different annotation styles with part segmentation, enabling effective use of coarse labels for improved performance.

Findings

Outperforms multi-task and semi-supervised baselines

Develops a new benchmark on Caltech-UCSD birds and OID Aircraft datasets

Effectively utilizes diverse coarse supervision signals

Abstract

A significant bottleneck in training deep networks for part segmentation is the cost of obtaining detailed annotations. We propose a framework to exploit coarse labels such as figure-ground masks and keypoint locations that are readily available for some categories to improve part segmentation models. A key challenge is that these annotations were collected for different tasks and with different labeling styles and cannot be readily mapped to the part labels. To this end, we propose to jointly learn the dependencies between labeling styles and the part segmentation model, allowing us to utilize supervision from diverse labels. To evaluate our approach we develop a benchmark on the Caltech-UCSD birds and OID Aircraft dataset. Our approach outperforms baselines based on multi-task learning, semi-supervised learning, and competitive methods relying on loss functions manually designed to exploit sparse-supervision.