Adapting AI-based speed violation detection systems for Africa: A case study with Nigeria

Abstract

Road traffic accidents pose a severe threat to public health and economic stability in Africa, with Nigeria bearing one of the highest road injury death rates globally. Speeding remains a significant contributor to fatal accidents across the continent, necessitating urgent interventions for effective speed management. In response, this paper presents a pioneering Speed Violation Detection System for Nigerian roads and a robust approach to Speed Violation Detection in Africa, leveraging artificial intelligence techniques.

Through targeted redesign and rigorous adaptation of existing AI-based speed detection strategies, this system introduces several modifications tailored for Nigerian road conditions. These include a custom tracking algorithm that eliminates the need for manual pixel-per-meter estimations, an enhanced vehicle detection model optimized for poor-quality road camera footage and unstructured traffic, and a localized license plate recognition system fine-tuned with a dataset of 24,242 Nigerian plate images. Additionally, the system integrates cloud-based data logging via AWS S3, ensuring secure, remote access for law enforcement agencies, thereby improving record-keeping and enforcement efficiency.

The system demonstrates strong capabilities in real-time vehicle detection, tracking, speed estimation, violators' plate-number capturing, and comprehensive data archiving. Its deployment holds promise for transforming road safety enforcement in Africa, offering law enforcement agencies invaluable tools to combat speeding and prevent accidents. The scalability and adaptability of the system underscore its potential for broader applications within Nigeria's traffic management infrastructure, marking a significant step towards achieving sustainable development goals centered on road safety and public well-being.