What question did this study set out to answer?

The research aims to develop a high-performance computer vision framework for automated traffic monitoring.

April 15, 2026Open Access

Neural ITS: A Real-Time Pipeline for High-Fidelity Traffic Analytics

Key Points

The research aims to develop a high-performance computer vision framework for automated traffic monitoring.
Utilized YOLOv8 for object detection.
Implemented DeepSORT for multi-object tracking.
Designed a 'Virtual Gate' for accurate vehicle counting in traffic scenarios.
Optimized for real-time data processing at over 30 frames per second.
Structured for integration with Smart City developments.
Achieved 96.5% mean average precision in vehicle detection and classification.
Demonstrated real-time performance exceeding 30 FPS on suitable infrastructure.
Successfully eliminated duplicate vehicle counts through innovative counting logic.

Abstract

Description Title: Neural ITS: A Real-Time Pipeline for High-Fidelity Traffic Analytics Author: Dr. Mustafa Omer Shoieb Mohammed Overview: This research presents a high-performance, non-invasive computer vision framework for automated traffic monitoring. The pipeline integrates YOLOv8 for state-of-the-art object detection with DeepSORT for robust multi-object tracking (MOT). Technical Highlights: Precision: Achieved 96.5% mAP in vehicle classification and detection. Performance: Optimized for real-time deployment, exceeding 30 FPS on RTSP-enabled infrastructure. Innovation: Features a "Virtual Gate" spatial counting logic to eliminate duplicate counts and ensure granular flow quantification. Scalability: Specifically architected for Smart City integration and AI-driven infrastructure management.

Neural ITS: A Real-Time Pipeline for High-Fidelity Traffic Analytics

Key Points

Abstract

Cite This Study