Robust Pedestrian Detection via Constructing Versatile Pedestrian Knowledge Bank

TL;DR

This paper introduces a versatile pedestrian knowledge bank derived from large-scale pretrained models to improve pedestrian detection across various scenes and frameworks, outperforming existing methods.

Contribution

It proposes a novel pedestrian knowledge bank that generalizes across scenes and detection frameworks, enhancing detection performance.

Findings

Outperforms state-of-the-art detection methods

Demonstrates high versatility across diverse scenes

Effectively enhances pedestrian feature representations

Abstract

Pedestrian detection is a crucial field of computer vision research which can be adopted in various real-world applications (e.g., self-driving systems). However, despite noticeable evolution of pedestrian detection, pedestrian representations learned within a detection framework are usually limited to particular scene data in which they were trained. Therefore, in this paper, we propose a novel approach to construct versatile pedestrian knowledge bank containing representative pedestrian knowledge which can be applicable to various detection frameworks and adopted in diverse scenes. We extract generalized pedestrian knowledge from a large-scale pretrained model, and we curate them by quantizing most representative features and guiding them to be distinguishable from background scenes. Finally, we construct versatile pedestrian knowledge bank which is composed of such representations, and then we leverage it to complement and enhance pedestrian features within a pedestrian detection framework. Through comprehensive experiments, we validate the effectiveness of our method, demonstrating its versatility and outperforming state-of-the-art detection performances.