Do Crash Barriers and Fences Have an Impact on Wildlife–Vehicle Collisions?—An Artificial Intelligence and GIS-Based Analysis

摘要: Wildlife–vehicle collisions (WVCs) cause significant road mortality of wildlife and have led to the installation protective measures along streets. Until now, it has been difficult determine impact roadside infrastructure that might act as a barrier for animals. The main deficits are lack geodata georeferenced accidents recorded larger area. We analyzed 113 km network district Freyung-Grafenau, Germany, 1571 WVCs, examining correlations between appearance presence or absence infrastructure, particularly crash barriers fences, relevance blocking effect individual species. To receive data on scale, we 5596 inspection images with neural recognition GIS complete spatial inventory. This was combined WVCs in evaluate infrastructure’s accidents. results show an lower roads barriers. In particular, smaller animals collision share. risk reduction at fenced sections could not be proven only available 3% roads. Thus, especially fence dataset must validated by sample number. However, these preliminary indicate combination artificial intelligence may used analyze better allocate apply alternative measures, such dynamic WVC risk-warning.