Advanced ROI (Region of Interest)

This project leverages the YOLOv5 and YOLOv10 pretrained models to implement interactive object detection with customizable polygonal Regions of Interest (ROIs). Users can define precise boundaries to enhance object detection in targeted areas, making it ideal for surveillance, traffic monitoring, and crowd management applications. Additionally, it features zone entry and exit detection for advanced video analytics, enabling effective tracking and monitoring in dynamic environments.

Key Features

• YOLOv5 and YOLOv10 Integration
• Customizable Polygonal ROI Boundaries
• Zone Entry and Exit Detection
• Real-Time Object Detection
• Interactive ROI Definition
• Enhanced Video Analytics
• Multi-Platform Compatibility
• Airborne Surveillance Optimization
• User-Friendly Interface
• High Detection Accuracy in Complex Environments

Model Overview

Basic Model	YOLOV5

YOLO Models

YOLOV5	YOLOV10

Performance Comparison

Parameters	Basic Model	ROI Model
Total Frame	2527	2527
Total Time	42.99 seconds	41.61 seconds
Average Inference Time	0.0049	0.0049
Average FPS	61.7	66.6

Our study on Advanced Region of Interest Detection using the YOLO Model with Mouse Interaction and Polygon Area Function reveals an 8% increase in FPS with the ROI model compared to the basic model. Additionally, the ROI model demonstrates more efficient CPU, GPU, and resource utilization, highlighting its practical advantage in real-world scenarios