One-Shot Learning for Semantic Segmentation

TL;DR

This paper introduces a one-shot learning approach for semantic segmentation, enabling dense pixel-level predictions for new classes with minimal annotated data, achieving significant accuracy improvements and faster inference.

Contribution

It extends low-shot learning techniques to dense segmentation by training a network that generates FCN parameters from few annotated images, improving accuracy and speed.

Findings

25% relative meanIoU improvement over baselines

At least 3 times faster inference

Effective on unseen classes in PASCAL VOC 2012

Abstract

Low-shot learning methods for image classification support learning from sparse data. We extend these techniques to support dense semantic image segmentation. Specifically, we train a network that, given a small set of annotated images, produces parameters for a Fully Convolutional Network (FCN). We use this FCN to perform dense pixel-level prediction on a test image for the new semantic class. Our architecture shows a 25% relative meanIoU improvement compared to the best baseline methods for one-shot segmentation on unseen classes in the PASCAL VOC 2012 dataset and is at least 3 times faster.