Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{196887,
        author = {Mrs. K Naga Maha Lakshmi and Goli Shivaprasad Reddy and Kota Prathiba and Dhanavath Manesh and Kempu Bharath},
        title = {An Intelligent Vision System for Predicting Pedestrian Behaviour in Traffic Scenes},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {4380-4386},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196887},
        abstract = {Pedestrian safety is a critical challenge in intelligent transportation systems due to the unpredictable nature of human movement in dynamic traffic environments. Existing systems primarily focus on detection and tracking, lacking the ability to predict future pedestrian behaviour, which is essential for proactive decision-making. This paper proposes STP-Vision, a spatio-temporal deep learning framework designed to detect, track, and predict pedestrian behaviour using real-time video data. The system integrates multiple components into a unified pipeline. Pedestrians are detected using a YOLO-based object detection model and tracked across frames using DeepSORT to maintain identity consistency. Motion features such as velocity, direction, and trajectory history are extracted and encoded into temporal sequences. A Long Short-Term Memory (LSTM) network is employed to predict future pedestrian positions based on historical movement patterns. A risk estimation module further evaluates potential collision scenarios using Time-to-Collision (TTC) metrics. Experimental evaluation demonstrates that the system achieves a detection accuracy of 94.2%, tracking accuracy of 90.5%, and trajectory prediction accuracy of approximately 89%. The system operates at 40 frames per second with an average latency of 300 ms, enabling real-time deployment. The proposed framework enhances situational awareness and contributes to safer autonomous driving systems.},
        keywords = {Pedestrian Behaviour Prediction, YOLO, DeepSORT, LSTM, Trajectory Prediction, Intelligent Transportation Systems.},
        month = {April},
        }