Poacher Activity Detection Device for Wildlife Conservation

Abstract

The scale of the wilderness makes it extremely difficult to constantly patrol for poaching activity. Consequently, there is an urgent need for an effective automated tool to curb poaching. The goal of this project was to build a low-cost device that can be installed in wilderness areas as well as on surveillance drones and can send real-time alerts of poaching activity to park officials. The device hardware consists of a Raspberry Pi computer with a power bank, camera, and an antenna. The test data mainly consisted of publicly available images of wild animals and humans in the wilderness. An image classification algorithm was built using a set of pre-trained AI models to detect the presence of humans, weapons, and animals in a photo; this algorithm was implemented on a device that can be used to notify officials of possible poaching activity. The device successfully meets all engineering goals. The design costs about US$100, takes about 5 minutes to run per image, is over 90% accurate in recognizing humans, weapons, and animals in all the test images, and can generate notifications. The poacher activity detection device may help rangers monitor large areas and decrease poaching activity.