Object Discovery with a Copy-Pasting GAN

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.

References in corpus (5)

• Semantic Instance Segmentation with a Discriminative Loss Function
• Semantic Instance Segmentation via Deep Metric Learning
• MONet: Unsupervised Scene Decomposition and Representation
• Unsupervised Cipher Cracking Using Discrete GANs
• TF-Replicator: Distributed Machine Learning for Researchers

Cited by in corpus (7)

• On the Binding Problem in Artificial Neural Networks
• Finding an Unsupervised Image Segmenter in Each of Your Deep Generative Models
• Object-aware Contrastive Learning for Debiased Scene Representation
• Multiple-Identity Image Attacks Against Face-based Identity Verification
• Learning to Manipulate Individual Objects in an Image
• Self-supervised Segmentation via Background Inpainting
• Disassembling Object Representations without Labels