Smart Wildlife Sentinel (SWS): Preventing Wildlife-Vehicle Collisions and Monitoring Road Ecology with Embedded IoT Systems and Machine Learning

Every year in the US, millions of animals are hit by vehicles, making wildlife vehicle collisions (WVCs) a real danger to both animal and human life. In addition, road networks are abiotic barriers to wildlife migration between regions, creating ripple effects on ecosystems. In this paper, a smart wildlife sentinel sign system is demonstrated, utilizing the technologies of Internet of Things (IoT), image recognition, data processing and visualization. This smart system is intended to prevent WVCs by warning driver early once sensors are triggered. Simultaneously, animal images are captured via infrared camera. Data processing and collection are conducted through computer vision filtering, convolutional neural network (CNN) based image classification, and animal activity metadata generation. Wildlife activity data can be exported wirelessly to cloud databases to assist ecologists and government road agencies to further analyze wildlife activity hotspots and potential migration patterns over time.