Bird Species Classification Using MobileNetV2

This paper proposes to develop and evaluate a machine-learning model for bird species classification using the MobileNetV2 architecture. The model is trained on a dataset of images of various bird species and evaluated on a separate test arranged to assess its precision in categorizing unseen pictures of birds. Transfer learning is utilized to further refine the pre-established model to the specific task of bird species classification. The model is estimated using criteria similar to delicacy, recall, perfection, and F1-score. The MobileNetV2 model that was recommended achieved an overall test set accuracy of 90.5%. Further experimentation with transfer learning techniques led to an accuracy improvement of 94.3%. The results demonstrate the feasibility of using MobileNetV2 for efficient and accurate bird species classification, with potential applications in fields such as ornithology and conservation. The model's contribution towards bird identification can aid in conservation efforts through improved population monitoring and habitat management.