
Deep Learning based Real Time Vehicle Detection and Collision Prediction with Distance-Based Alert System
Deep Learning based Real Time Vehicle Detection and Collision Prediction with Distance-Based Alert System
IEEE BASE PAPER TITLE:
Collision Prediction in an Integrated Framework of Scenario-Based and Data-Driven Approaches
IEEE BASE PAPER ABSTRACT
A collision prediction framework integrating scenario-based approach with data-driven approach is proposed to enhance the safety of autonomous driving vehicles as well as advanced driver assistance systems. No matter how the autonomous driving is intelligent, it is inevitable to consider malfunction or faults of sensors, actuators, and processors, thus resulting in the collision. To address these issues, several studies have been proposed to improve performance based on model-based or data-driven approaches. However, there are several challenges in terms of the scarcity of accident data and the lack of explainability of deep neural networks. To overcome the limits of both approaches, an integrated framework that includes trajectory prediction, threat assessment, and decision-making based on convolutional neural network (CNN) for collision prediction is introduced. For more detail, both trajectory prediction based on Kalman filter and probabilistic threat metric are added in the form of a simplified bird’s eye view (SBEV), which is the input to the network. In the development of the proposed algorithm, pre-crash simulation data and experimental data have been employed. A comparative study shows that the proposed algorithm outperforms the model-based algorithm on simulation data containing safety-critical scenarios. Furthermore, it outperforms the data-driven algorithm on experimental data.
PROJECT OUTPUT VIDEO:
ALGORITHM / MODEL USED:
Pre-trained YOLOv4 Model.
OUR PROPOSED ABSTRACT:
Road accidents remain a significant global concern, causing extensive loss of life and property every year. With the increasing volume of vehicles on the road, the need for advanced systems to prevent collisions and ensure road safety has become paramount. Addressing this critical challenge, the project “Deep Learning-Based Real-Time Vehicle Detection and Collision Prediction with Distance-Based Alert System” aims to utilize modern technologies to reduce the risk of road accidents through timely detection and alerts.
Developed using MATLAB, the system integrates the YOLOv4 (You Only Look Once version 4) model for real-time vehicle detection, trained on the COCO (Common Objects in Context) dataset. By analyzing testing videos sourced from a public database, the system accurately identifies vehicles within the scenes and measures their proximity to the camera-mounted vehicle.
Through a combination of distance measurement algorithms and alert generation mechanisms, drivers are promptly notified of potential collision risks via text and sound alerts. Finally, the Alert System issues timely warnings based on the calculated distances. Alerts are categorized into three levels: “Vehicle Very Near” for immediate collision risk, “Vehicle Little Far” for potential risk, and “Vehicle Very Far” for a safe distance, encouraging continued vigilance. These alerts are conveyed through visual and auditory signals, ensuring drivers are informed in real time to take preventive action.
This proactive system aims to enhance road safety by mitigating the occurrence of accidents and ultimately saving lives on the road. Experimental evaluations underscore the effectiveness and efficiency of the proposed approach, positioning it as a promising solution for integration into intelligent transportation systems. This project addresses the urgent need to mitigate road accidents by combining deep learning capabilities with precise distance-based alerts. By enabling real-time vehicle detection and collision prediction, it offers a scalable, effective, and practical solution to enhance road safety and save lives.
OBJECTIVES
- Develop a robust and efficient deep learning-based system for real-time vehicle detection in traffic scenes using the YOLOv4 model, trained on the COCO dataset, to accurately identify and localize vehicles within the camera’s field of view.
- Implement a distance measurement algorithm to assess the proximity between detected vehicles and the camera-mounted vehicle, converting pixel distances into real-world measurements to accurately gauge collision risks and provide timely warnings to drivers.
- Design and integrate an alert generation mechanism that utilizes both textual and auditory cues to notify drivers of potential collision risks based on the calculated distances, ensuring prompt and effective responses to prevent accidents and enhance road safety.
SYSTEM REQUIREMENTS:
HARDWARE REQUIREMENTS:
- System : Pentium i3 Processor.
- Hard Disk : 500 GB.
- Monitor : 15’’ LED.
- Input Devices : Keyboard, Mouse.
- Ram : 8 GB.
SOFTWARE REQUIREMENTS:
- Operating system : Windows 10 / 11.
- Coding Language : MATLAB
- Tool : MATLABR2024B
REFERENCE:
SUNGWOO LEE, BONGSOB SONG, AND JANGHO SHIN, “Collision Prediction in an Integrated Framework of Scenario-Based and Data-Driven Approaches”, in IEEE Access, vol. 12, pp. 55234-55247, 2024.