Vision-Based Pedestrian's Crossing Risky Behavior Extraction and Analysis for Intelligent Mobility Safety System

Area of Science:

• Traffic Safety
• Computer Vision
• Behavioral Analysis

Background:

• Crosswalks pose significant risks to pedestrians due to a lack of detailed behavioral data.
• Analyzing near-miss collisions and risky behaviors is crucial for enhancing road safety infrastructure.

Purpose of the Study:

• To develop a novel approach for extracting risky behaviors of vehicles and pedestrians from CCTV footage.
• To analyze interactive moving patterns within crossing environments using extracted behavioral features.

Main Methods:

• Recasting surveillance CCTV cameras for detailed analysis of the crossing environment.
• Implementing a sequential video processing pipeline for behavioral feature extraction.
• Analyzing extracted features to understand road user interactions and risky behaviors.

Main Results:

• Demonstrated a feasible method for obtaining potential risky behaviors from road-based CCTV cameras.
• Successfully extracted and analyzed behavioral features and interactive patterns at crosswalks.
• Validated the model using diverse video footage from Osan City, Republic of Korea.

Conclusions:

• The developed approach provides a foundation for understanding road user risky behaviors.
• The findings support decision-makers in implementing effective strategies for safer road environments.
• Utilizing CCTV data offers a valuable tool for proactive traffic safety improvements.