From Pedestrian Detection to Crosswalk Estimation: An EM Algorithm and Analysis on Diverse Datasets

TL;DR

This paper introduces an EM algorithm for estimating pedestrian crosswalk features from LiDAR and camera data, validated on diverse real-world datasets, with visualization tools provided.

Contribution

The work presents a novel EM algorithm for crosswalk estimation from sensor detections, applicable to both marked and unmarked crosswalks, with comprehensive analysis on multiple datasets.

Findings

Effective crosswalk parameter estimation demonstrated on real datasets

Algorithm performs well for both marked and unmarked crosswalks

Visualization tools aid in understanding pedestrian trajectories and crossing phases

Abstract

In this work, we contribute an EM algorithm for estimation of corner points and linear crossing segments for both marked and unmarked pedestrian crosswalks using the detections of pedestrians from processed LiDAR point clouds or camera images. We demonstrate the algorithmic performance by analyzing three real-world datasets containing multiple periods of data collection for four-corner and two-corner intersections with marked and unmarked crosswalks. Additionally, we include a Python video tool to visualize the crossing parameter estimation, pedestrian trajectories, and phase intervals in our public source code.