84 Birds Classification Using Transfer Learning and EfficientNetB2

Nowadays, global warming, wildfires, and air pollution have caused significant effects and seriously threatened the life and diversity of animals in common and birds in particular. Therefore, the conservation of birds is very urgent. Identify and classify them for statistics on the number of species, and distribution, and as a way to come up with reasonable conservation measures from scientists who study the environment and animals. In this research, we propose a new approach to classify birds using the EfficientNetB2 model, transfer learning techniques, and customizing the model’s hyperparameters. Model trained on the original dataset with the highest accuracy of 93% on both the validation and the testing set.