Bird Species Classification Using MobileNetV2 and CNN: A Comparative Study with Web Deployment

Bird species classification plays a crucial role in ecological research, biodiversity monitoring, and wildlife conservation. In this study, we propose a machine learning-based image classification system for identifying bird species using a custom dataset of 20 bird species. We explore and compare the performance of a Convolutional Neural Network (CNN) and MobileNetV2 architecture for classification. With a total of 3408 images divided into training, validation, and test sets, we fine-tuned both models and found that MobileNetV2 significantly outperformed the custom CNN model in terms of accuracy and generalization. The trained MobileNetV2 model was then converted to TensorFlow Lite (.tflite) format and integrated into a Flask-based web application. The system predicts the species name with associated confidence or informs the user if the species is outside the known dataset. This research not only highlights the potential of deep learning in species classification but also demonstrates real-world application deployment