ABSTRACT
The natural habitat of India is under stress because of the problems faced by the human development like over populace, pollutants, forest fragmentation and so on. The understanding of forest environment, identification of numerous animal species, quantifying the animal species and studying the behaviour of animal species are the significant areas to save the natural animal species. The manual process of identification of animals is a time-ingesting challenge and complicated in nature because of time obstacle and availability of suitable human beings. In this work, an automated identification of animal species is proposed under two phases using image processing techniques and Convolution Neural Network (CNN) based approaches, respectively. The CNN approach can classify the images of the target species and the image processing techniques in terms of pre-processing and feature extraction to successfully achieve the goals. The first phase is implemented using image pre-processing, RGB image to grey conversion, binary conversion, feature extraction and feature reduction. The second phase has the design of CNN, training CNN and testing CNN. The CNN is supplied with the extracted features and targeted animal species are identified with a classification accuracy of 90%.
