Dense Teacher: Dense Pseudo-Labels for Semi-supervised Object Detection

TL;DR

This paper introduces Dense Teacher, a semi-supervised object detection method using dense pseudo-labels that eliminate the need for post-processing, resulting in improved performance over traditional pseudo-box approaches.

Contribution

The paper proposes Dense Pseudo-Labels (DPL) as a new form of pseudo-labels that are dense and do not require post-processing, enhancing semi-supervised object detection.

Findings

Outperforms pseudo-box methods on COCO and VOC datasets

Eliminates the need for post-processing in pseudo-label generation

Demonstrates robustness across various semi-supervised settings

Abstract

To date, the most powerful semi-supervised object detectors (SS-OD) are based on pseudo-boxes, which need a sequence of post-processing with fine-tuned hyper-parameters. In this work, we propose replacing the sparse pseudo-boxes with the dense prediction as a united and straightforward form of pseudo-label. Compared to the pseudo-boxes, our Dense Pseudo-Label (DPL) does not involve any post-processing method, thus retaining richer information. We also introduce a region selection technique to highlight the key information while suppressing the noise carried by dense labels. We name our proposed SS-OD algorithm that leverages the DPL as Dense Teacher. On COCO and VOC, Dense Teacher shows superior performance under various settings compared with the pseudo-box-based methods.