MVUDA: Unsupervised Domain Adaptation for Multi-view Pedestrian Detection

TL;DR

This paper introduces an unsupervised domain adaptation method for multi-view pedestrian detection that adapts models to new camera setups without additional labeled data, improving robustness and deployment flexibility.

Contribution

It proposes a novel UDA approach using mean teacher self-training with tailored pseudo-labeling for multi-view pedestrian detection, reducing reliance on labeled monocular datasets.

Findings

Achieves state-of-the-art results on benchmarks like MultiviewX to Wildtrack.

Effectively adapts models to new camera rigs without labeled data.

Validates key design choices through extensive evaluations.

Abstract

We address multi-view pedestrian detection in a setting where labeled data is collected using a multi-camera setup different from the one used for testing. While recent multi-view pedestrian detectors perform well on the camera rig used for training, their performance declines when applied to a different setup. To facilitate seamless deployment across varied camera rigs, we propose an unsupervised domain adaptation (UDA) method that adapts the model to new rigs without requiring additional labeled data. Specifically, we leverage the mean teacher self-training framework with a novel pseudo-labeling technique tailored to multi-view pedestrian detection. This method achieves state-of-the-art performance on multiple benchmarks, including MultiviewX$\rightarrow$Wildtrack. Unlike previous methods, our approach eliminates the need for external labeled monocular datasets, thereby reducing reliance on labeled data. Extensive evaluations demonstrate the effectiveness of our method and validate key design choices. By enabling robust adaptation across camera setups, our work enhances the practicality of multi-view pedestrian detectors and establishes a strong UDA baseline for future research.