Object Discovery with a Copy-Pasting GAN

TL;DR

This paper introduces a novel unsupervised object discovery method using a copy-pasting GAN, where the generator learns to identify and segment objects by fooling the discriminator through compositing objects into different images.

Contribution

A new copy-pasting GAN framework for unsupervised object discovery that learns to segment objects without any direct supervision.

Findings

Works well on four diverse datasets

Handles large object variations and cluttered backgrounds

Effective in unsupervised object segmentation

Abstract

We tackle the problem of object discovery, where objects are segmented for a given input image, and the system is trained without using any direct supervision whatsoever. A novel copy-pasting GAN framework is proposed, where the generator learns to discover an object in one image by compositing it into another image such that the discriminator cannot tell that the resulting image is fake. After carefully addressing subtle issues, such as preventing the generator from `cheating', this game results in the generator learning to select objects, as copy-pasting objects is most likely to fool the discriminator. The system is shown to work well on four very different datasets, including large object appearance variations in challenging cluttered backgrounds.