Abstract:Pedestrian detection is used in many vision based applications ranging from video surveillance to autonomous driving. Despite achieving high performance, it is still largely unknown how well existing detectors generalize to unseen data. This is important because a practical detector should be ready to use in various scenarios in applications. To this end, we conduct a comprehensive study in this paper, using a general principle of direct cross-dataset evaluation. Through this study, we find that existing state-of-the-art pedestrian detectors, though perform quite well when trained and tested on the same dataset, generalize poorly in cross dataset evaluation. We demonstrate that there are two reasons for this trend. Firstly, their designs (e.g. anchor settings) may be biased towards popular benchmarks in the traditional single-dataset training and test pipeline, but as a result largely limit their generalization capability. Secondly, the training source is generally not dense in pedestrians and diverse in scenarios. Under direct cross-dataset evaluation, surprisingly, we find that a general purpose object detector, without pedestrian-tailored adaptation in design, generalizes much better compared to existing state-of-the-art pedestrian detectors. Furthermore, we illustrate that diverse and dense datasets, collected by crawling the web, serve to be an efficient source of pre-training for pedestrian detection. Accordingly, we propose a progressive training pipeline and find that it works well for autonomous-driving oriented pedestrian detection. Consequently, the study conducted in this paper suggests that more emphasis should be put on cross-dataset evaluation for the future design of generalizable pedestrian detectors. Code and models can be accessed at <a class="link-external link-https" href="https://github.com/hasanirtiza/Pedestron" rel="external noopener nofollow">this https URL</a>.

Towards Rich, Portable, and Large-Scale Pedestrian Data Collection

TBD Pedestrian Data Collection: Towards Rich, Portable, and Large-Scale Natural Pedestrian Data

A Richly Annotated Dataset for Pedestrian Attribute Recognition

Generalizable Pedestrian Detection: The Elephant In The Room

Pedestrian Attribute Recognition: A New Benchmark Dataset and A Large Language Model Augmented Framework

The Pedestrian Patterns Dataset

A Richly Annotated Pedestrian Dataset for Person Retrieval in Real Surveillance Scenarios

When Pedestrian Detection Meets Multi-Modal Learning: Generalist Model and Benchmark Dataset

PSI: A Pedestrian Behavior Dataset for Socially Intelligent Autonomous Car

VersatileGait: A Large-Scale Synthetic Gait Dataset Towards in-the-Wild Simulation

Diving Deeper Into Pedestrian Behavior Understanding: Intention Estimation, Action Prediction, and Event Risk Assessment

A Large-Scale Synthetic Gait Dataset Towards In-the-wild Simulation and Comparison Study.

Data-Driven Reachability Analysis of Pedestrians Using Behavior Modes

DELTA: Integrating Multimodal Sensing with Micromobility for Enhanced Sidewalk and Pedestrian Route Understanding

Gait Recognition in Large-scale Free Environment via Single LiDAR

Top-view Trajectories: A Pedestrian Dataset of Vehicle-Crowd Interaction from Controlled Experiments and Crowded Campus

APE: An Open and Shared Annotated Dataset for Learning Urban Pedestrian Path Networks

Seek and You Will Find: A New Optimized Framework for Efficient Detection of Pedestrian

Graph-SIM: A Graph-based Spatiotemporal Interaction Modelling for Pedestrian Action Prediction

A novel approach for reliable pedestrian trajectory collection with behavior-based trajectory reconstruction for urban surveillance systems

Pedestrian Detection: Domain Generalization, CNNs, Transformers and Beyond