An improved safety voilation detection algorithm based on DeepLearning

Project Code :TCPGPY2067

Objective

The primary goal of this project is to develop an automated system for detecting traffic violations through advanced deep learning models, specifically YOLOv8, YOLO26 and YOLOv11. This system will be capable of detecting and classifying a range of traffic violations, including not wearing helmets, triple riding, phone usage while riding, and wheeling. The project will involve the development of a Python-based backend to process real-time video feeds, ensuring efficient detection of violations as they occur. Additionally, a user-friendly front-end interface will be designed using Streamlit, enabling traffic authorities to easily visualize the results of the detection system and interact with the data. Ultimately, the system aims to improve road safety by providing a reliable, automated solution for real-time traffic violation detection, making the process faster, more accurate, and scalable.

Abstract

Traffic safety violations in recent years have created major obstacles for managing road safety. The traditional systems which track traffic violations need manual work to operate but this approach fails to provide effective results during live situations. The research paper presents a solution for detecting safety violations which include "Not Wearing Helmet," "Triple Riding," "Usage of Phone While Riding," and "Wheeling" through deep learning algorithms. The research study aims to enhance safety violation detection results through the implementation of both YOLOv26, YOLO11 and YOLOv8 detection systems. The existing systems face challenges with real-time violation detection because they cannot handle the unpredictable conditions of active traffic situations. The system uses advanced object detection systems which were developed to identify objects in a collection of traffic images that have been labeled. The comparative assessment demonstrates that the proposed system delivers better performance than both standard approaches and current deep learning technologies. The system which this paper proposes creates a strong solution for monitoring traffic in real-time while detecting violations thus promoting safer road conditions.

Keywords: YOLOv8, YOLOv11,YOLO26, helmet violation, triple riding, phone usage detection, wheeling, deep learning, object detection, traffic safety, real-time monitoring.

NOTE: Without the concern of our team, please don't submit to the college. This Abstract varies based on student requirements.

Block Diagram

Specifications

4.2 Hardware Requirements

Processor: Intel i3 or equivalent for handling basic video processing.
RAM: 8 GB for smooth operation.
Storage: 160 GB hard disk.
Cameras:

High-definition CCTV cameras (1080p or higher resolution).
Frame rate: Minimum 30 frames per second (FPS).

GPU: None specified, but NVIDIA GTX 1080 or higher is recommended for faster inference in deep learning tasks.
Network: High-speed internet connection (minimum 1 Gbps) for smooth data transfer.

4.3 Software Requirements

Operating System:

Windows 7/8/10 for the front-end development.
Linux-based OS (Ubuntu 18.04 or higher) for backend deployment.

Server-side Script: HTML, CSS, Bootstrap, and JS.
Programming Languages:

Python (for backend and model training).
JavaScript, HTML, CSS (for front-end development via Streamlit).

Deep Learning Frameworks:

TensorFlow or PyTorch (for YOLOv8, v11, model implementation).
OpenCV (for video processing and frame extraction).

Libraries and Tools:

Streamlit (for the front-end interface).
Flask (for API integration between backend and front-end).
NumPy, Pandas (for data processing and analysis).

Database: MySQL or PostgreSQL (for storing logs and metadata of violations).
Version Control: Git for version control and collaboration; GitHub or GitLab for repository hosting and CI/CD pipelines.
IDE/Workbench: PyCharm or VSCode for Python development.
Server Deployment: XAMPP Server.