Wild Animal Detection from Highly Cluttered Forest Images Using Deep Residual Networks

摘要

Wild animal detection is a dynamic research field since last decades. The videos acquired from camera-trap comprises of scenes that are cluttered that poses a challenge for detection of the wild animal. In this paper, we proposed a deep learning based system to detect wild animal from highly cluttered natural forest images. We have utilized Deep Residual Network (ResNet) for features extraction from cluttered forest images. These features are feed to classification through some of the best in class machine learning techniques, to be specific Support Vector Machine, K-Nearest Neighbor and Ensemble Tree. Our outcomes demonstrate that our detection system through ResNet outperforms compare to existing systems reported in the literature.