HUWSOD: Holistic Self-training for Unified Weakly Supervised Object Detection

TL;DR

HUWSOD introduces a unified self-training framework for weakly supervised object detection that replaces traditional proposals with self-supervised modules, achieving competitive results without external supervision.

Contribution

The paper presents HUWSOD, a novel end-to-end WSOD network utilizing self-supervised proposal generation and holistic self-training, eliminating the need for external modules or additional supervision.

Findings

HUWSOD achieves performance close to fully-supervised Faster R-CNN.

Self-supervised proposal generators outperform traditional object proposals.

Randomly initialized boxes are effective for WSOD training.

Abstract

Most WSOD methods rely on traditional object proposals to generate candidate regions and are confronted with unstable training, which easily gets stuck in a poor local optimum. In this paper, we introduce a unified, high-capacity weakly supervised object detection (WSOD) network called HUWSOD, which utilizes a comprehensive self-training framework without needing external modules or additional supervision. HUWSOD innovatively incorporates a self-supervised proposal generator and an autoencoder proposal generator with a multi-rate resampling pyramid to replace traditional object proposals, enabling end-to-end WSOD training and inference. Additionally, we implement a holistic self-training scheme that refines detection scores and coordinates through step-wise entropy minimization and consistency-constraint regularization, ensuring consistent predictions across stochastic augmentations of the same image. Extensive experiments on PASCAL VOC and MS COCO demonstrate that HUWSOD competes with state-of-the-art WSOD methods, eliminating the need for offline proposals and additional data. The peak performance of HUWSOD approaches that of fully-supervised Faster R-CNN. Our findings also indicate that randomly initialized boxes, although significantly different from well-designed offline object proposals, are effective for WSOD training.