Monitoring Animals on the Road to Reduce Roadkill

Roadkill, which is animals that got killed by Animal-Vehicle Collisions, has turned out to be a global problem. However, a panacea is still at the experimental level. AVCs are a threat to biodiversity, human safety and the economy. Hence, mitigating AVCs is in the best interest of animals as well as humans.

AVCs are not randomly occurring, not uniformly distributed, and cannot be treated equally for all species. In this context, it is essential to identify spatiotemporal AVC hotspots concerning each species to implement tailored mitigation strategies. Nonetheless, existing roadkill hotspot identification strategies are inherent with limitations.

Prevailing approaches are focused only on carcasses and they are biased toward large animals like kangaroos and koalas. Further, strategies that depend on voluntary citizen scientists are faced with long-term challenges, such as the need to reward and motivate participants.
These limitations and challenges have contributed to building an inconsistent, unreliable, inaccurate, and incomplete picture of spatiotemporal hotspots of small species like turtles.
Therefore, we have identified a demand for an innovative approach utilising computer vision techniques to pinpoint small animals on the roads and roadsides, which mitigates roadkill by allowing unbiased identification of spatiotemporal small animal hotspots without the active involvement of road users.

This study has a significant contribution to the application area of road ecology and the computer vision discipline area. Firstly, it will improve the ecological survey results of roadkill incidents on road networks operated by organisations and stakeholders such as Transport for New South Wales (NSW), local councils and other road infrastructure operators. Based on these improved results, the above stakeholders can implement cost-effective design of potential mitigation measures to minimise AVCs. This will lead to improved road safety and reduced adverse effects on the animal population in Australia. Moreover, it will contribute to the 11th and 15th Sustainable Development Goals (SDGs) of the United Nations by ensuring sustainable transport and protecting life on land. Secondly, the enhanced object detection algorithms to detect small objects in fast-moving videos will enable applications like self-driving cars and unmanned aerial vehicles to identify obstacles at extended distances.