An Intelligent Crowd Density and Motion Direction Estimation for Real-Time Crowd Dynamics
Sarbartha Sankar Mallick, Indrajit Das, S. Jerome Das, Raja Karmakar · 2024
Efficient crowd management is essential for ensuring safety and security in various societal gatherings, such as public events and transportation hubs. Traditional crowd surveillance methods often fall short in handling the complexities inherent in large-scale gatherings. To address these challenges, this paper explores the application of computer vision techniques for crowd density estimation and motion direction analysis. The proposed approach investigates the integration of advanced computer vision algorithms to monitor crowd dynamics in real time. Leveraging state-of-the-art techniques like YOLOv8 for object detection, for optical flow estimation, we aim to achieve unparalleled accuracy in crowd analysis. We use the Lucas-Kanade optical flow combining with corner to track moving target. We use a Convolutional Neural Network (CNN) to extract local features in the preprocessed video. These features are sequenced utilizing a Long Short Term Memory (LSTM). The results of the proposed approach showcase the effectiveness of the proposed approach in accurately estimating crowd density and analyzing motion direction across diverse scenarios. Using computer vision, our mechanism provides timely insights into crowd behavior, enabling proactive interventions to prevent potential incidents and ensure public safety. This work underscores the pivotal role of advanced computer vision techniques in revolutionizing crowd management strategies.