Image analysis using convolutional neural network to detect bird species

Abstract

Recognition for bird species is one of the most challenging tasks in visualizing a very abundant bird species. It is a subfield of object recognition and is meant to differentiate sub-categories. Furthermore, it aims to categorize the data, such as images. Therefore, identifying the bird species through images helps us acquire a more accurate result. Thus, we have developed a deep learning model to help users identify bird species. Meanwhile, deep convolutional neural networks have been very effective in dealing with images of having different dimensions and thus succeeding at tasks like object detection and classification, learning various features like edges, colors, etc. Thus, using the neural network model, we have taken 200 different species of birds in the form of images to classify. First, birds images were learned by a convolutional neural network (CNN) to identify the dominant features in the images. Then, we used the softmax function to obtain a probability distribution of bird characteristics. Finally, to obtain a more accurate result, we compare our GoogleLeNet model with the VGG16.