Virtual Booth

Animal–vehicle collisions remain a serious safety issue, especially on rural and semi-rural roads in Malaysia, where animals such as cats, dogs, and cows frequently cross roadways. These incidents often occur at night and can result in injuries, vehicle damage, and the loss of animal life. Although modern driver assistance and autonomous driving systems are well developed for detecting vehicles and pedestrians, the detection of small animals is still limited and often delayed. Statistics show that around 508 animal–vehicle collision cases were recorded in Malaysia in 2016, while reports from 2019 to 2024 indicate that at least 39 animals were struck by autonomous vehicles.

To address this problem, this research presents a camera-based animal detection system using deep learning to provide early warnings to drivers. The system is designed to work for both autonomous and non-autonomous vehicles, making it practical and cost-effective. A deep learning model based on the You Only Look Once (YOLO) algorithm was trained using annotated images of common Malaysian animals, namely cats, dogs, and cows. Among the evaluated models, YOLOv8 demonstrated the best performance in terms of detection accuracy and reliability.

The proposed system generates real-time visual and audio alerts, with the potential to integrate GPS-based warnings to inform nearby drivers of animal crossings. Overall, this project aims to improve road safety, reduce animal-related accidents, and offer a scalable solution that can support smarter and safer transportation systems in the future.