A Robust Anchor-based Method for Multi-Camera Pedestrian Localization

TL;DR

An anchor-based method that leverages fixed-position anchors to reduce the impact of camera parameter errors and significantly improves localization accuracy and remains resilient to noise in camera parameters, compared to methods without anchors.

Abstract

This paper addresses the problem of vision-based pedestrian localization, which estimates a pedestrian's location using images and camera parameters. In practice, however, calibrated camera parameters often deviate from the ground truth, leading to inaccuracies in localization. To address this issue, we propose an anchor-based method that leverages fixed-position anchors to reduce the impact of camera parameter errors. We provide a theoretical analysis that demonstrates the robustness of our approach. Experiments conducted on simulated, real-world, and public datasets show that our method significantly improves localization accuracy and remains resilient to noise in camera parameters, compared to methods without anchors.

Figures (11)

• Figure 1: An illustration of detection-based localization problem. Given the input images (labeled by detection boxes) and parameters of multiple cameras, we aim to recover the location of the targets in space.
• Figure 2: Comparison of no-anchor method and anchor-based method. The no-anchor localization method directly minimizes the reprojection error, whereas the anchor-based method minimizes the anchor-adjusted reprojection error.
• Figure 3: Left: The 2D map of simulation environment. The circles represent the locations of each camera while the corresponding lines are their orientations. Right: The simulated trajectories.
• Figure 4: Average Distance (m) of different localization methods under camera parameter errors in Table \ref{['tab: pix and ang error']}. The data point is the average of both positive and negative perturbations.
• Figure 5: Improvement Ratio of different localization methods under camera parameter errors in Table \ref{['tab: pix and ang error']}. The data point is the average of both positive and negative perturbations.