Bird Species Detection and Classification Based on HOG Feature Using Convolutional Neural Network

Abstract

Bird detection and classification for real time image are com-plex tasks but crucial improvement can be performed by the best computer vision algorithms to solve such kinds of problems. Some challenges like lighting conditions of the images, similarities in subspecies of birds may be raised. During the last few years, CNN appeared as the state of the art with regard to the accuracy for a number of computer vision work such as image classification, object detection, and segmentation. To identify the object like bird CNN has been used in this work. Here, we have designed bird detection and classification system based on Gaussian and Gabor filters, HOG as well as Convolutional Neural Networks. HOG is one of the accepted features for object detection and it can be extracted from all portions of the image. So, Histogram of Oriented Gradients (HOG) is used to implement the CNN. We have done experiment with the standard datasets and have found better results than other methods by applying above procedures.