An automated vision-based method to detect elephants for mitigation of human-elephant conflicts

Human-elephant conflict is a frontline conservation issue in the world. The loss and fragmentation of elephant's habitat owing to the increased human encroachment leads to a notable conservation issue. It raises the need for a non-invasive and efficient solution for the mitigation of human-elephant conflicts. Consequently, deploying a vision based surveillance method in the real time environment can prove to be significantly useful to provide the warnings well in advance thereby reducing the human elephant conflict. In this paper, a method for the identification of elephant as an object using image processing is proposed. The method dynamically learns from the trained images with different backgrounds, lighting conditions. Further, it classifies the input image based on the features of color and texture. The outcomes demonstrate that the proposed method effectively deals with the detection of elephants in near and far distances, cluttered and occluded environment.