PGTRNet: Two-phase Weakly Supervised Object Detection with Pseudo Ground Truth Refinement

TL;DR

PGTRNet introduces a two-phase weakly supervised object detection method that refines pseudo ground truth using multiple bounding boxes and online refinement, significantly improving detection accuracy without extra parameters.

Contribution

It proposes a novel pseudo ground truth refinement network that enhances pseudo label quality and decouples training phases, achieving state-of-the-art results in weakly supervised object detection.

Findings

Boosts mAP by 2.1% on PASCAL VOC 2007

Effectively refines pseudo ground truth during training

Achieves state-of-the-art performance in WSOD

Abstract

Current state-of-the-art weakly supervised object detection (WSOD) studies mainly follow a two-stage training strategy which integrates a fully supervised detector (FSD) with a pure WSOD model. There are two main problems hindering the performance of the two-phase WSOD approaches, i.e., insufficient learning problem and strict reliance between the FSD and the pseudo ground truth (PGT) generated by the WSOD model. This paper proposes pseudo ground truth refinement network (PGTRNet), a simple yet effective method without introducing any extra learnable parameters, to cope with these problems. PGTRNet utilizes multiple bounding boxes to establish the PGT, mitigating the insufficient learning problem. Besides, we propose a novel online PGT refinement approach to steadily improve the quality of PGT by fully taking advantage of the power of FSD during the second-phase training, decoupling the first and second-phase models. Elaborate experiments are conducted on the PASCAL VOC 2007 benchmark to verify the effectiveness of our methods. Experimental results demonstrate that PGTRNet boosts the backbone model by 2.1% mAP and achieves the state-of-the-art performance.