Species Identification of Oriental Magpie-robin Using Convolutional Neural Network in Bangladesh

Abstract

To understand the behavior of animals, monitoring the population of animals is necessity. Extinction is a common phenomenon nowadays. So, understanding and ecosystem is a must to preserve each species. The global climate change is a common issue which affecting human race. Change in nature directly affects ecosystem, and a gradual change in bird’s pattern is seen. Monitoring the movement and population of bird gives us any indication about changes in environment. We have used convolutional neural network (CNN) model to classify birds with better accuracy to identify Magpie Robin which is the national bird of Bangladesh. By the use of Image Rotation and Noise injection, our data is augmented. First we have feed our data into the CNN model for training. The stochastic gradient descent was used along with mini batch size 32. Our model has brought an accuracy of 91.44%. By comparing many other models, our model proved to be effective and productive in classification of bird based on image processing.