Animal Images Classification with MobileNetV2

This study presents a deep learning based approach to classify 55 different animal species from image data. The main goal of the model is to develop a system that can classify different animal species with low hardware requirements and high accuracy. This model aims to produce solutions for real-world problems. In this respect, it is a userfriendly and versatile work that goes beyond theoretical research and has practical implications. In the system, MobileNetV2 is used as the basic convolutional neural network (CNN) architecture. This architecture is particularly suitable for real-time applications and embedded systems with its lightweight structure and high accuracy. The model is trained with transfer learning technique to achieve higher performance with limited training data. Furthermore, extensive preprocessing and data augmentation methods are applied to improve its generalization capability. The results of this study show that the MobileNetV2-based model achieves high performance on a multi-class animal classification task.