Object Segmentation Without Labels with Large-Scale Generative Models

TL;DR

This paper demonstrates that large-scale unsupervised generative models, specifically GANs, can effectively perform object segmentation without any labeled data, achieving state-of-the-art results on standard benchmarks.

Contribution

It introduces a novel approach using unsupervised GANs for object segmentation, eliminating the need for pixel-level or image-level labels.

Findings

Outperforms existing unsupervised methods on standard benchmarks

Achieves high-quality saliency masks differentiating foreground and background

Sets new state-of-the-art in label-free object segmentation

Abstract

The recent rise of unsupervised and self-supervised learning has dramatically reduced the dependency on labeled data, providing effective image representations for transfer to downstream vision tasks. Furthermore, recent works employed these representations in a fully unsupervised setup for image classification, reducing the need for human labels on the fine-tuning stage as well. This work demonstrates that large-scale unsupervised models can also perform a more challenging object segmentation task, requiring neither pixel-level nor image-level labeling. Namely, we show that recent unsupervised GANs allow to differentiate between foreground/background pixels, providing high-quality saliency masks. By extensive comparison on standard benchmarks, we outperform existing unsupervised alternatives for object segmentation, achieving new state-of-the-art.