Bird Species Identification Using CNN

Abstract

In our world, there are above 11000 bird species. Some bird species are being found rarely. Bird identification is a challenging task that usually leads to unclear labelling. When presented a picture of a bird, even professional bird watchers differ on the species. Despite having the same basic components across all bird species, form and appearance can vary greatly. Intraclass variance is substantial due to variations in lighting and backdrop, as well as a wide range of instances. Additionally, visual recognition of birds by humans is more comprehensible than audible recognition of birds. Consequently, we utilised convolutional neural networks (CNN). CNNs are a powerful Deep Learning ensemble that have shown to be effective in image processing. The dataset is used for both training and testing of a CNN system that classifies bird species. Everyone can quickly determine the name of the bird they wish to know by following this strategy. Keywords— Bird species, Image Processing, Convolutional Neural Networks.