African Birds Identification Using Swing Transformer
B Senthilnayaki, Maram Y. Al-Safarini, Shanthi Thangam Manukumar, T Aswani
Abstract
The popular hobby of birdwatching necessitates the use of bird books to determine the name of the bird. The birdwatchers observe the birds with their naked eyes for detection and identification of birds. The existing method can be inaccurate. to provide birders a useful tool to appreciate the beauty of distinct types of birds. This proposed system uses a dataset of 5,700 images. With the help of this bird classification system, the standard image dataset is converted to grayscale, and the pixel normalisation technique is applied to pre-processing the bird's dataset. Using software based on the idea of image recognition, SWING transformer is used for categorization to assist users in identifying different bird species. The accuracy of the system generated by this proposed model is 91%.