Enhancing Mammal Conservation Through Automated Species Classification Using Deep Learning

The decline in mammal populations is an urgent concern driven by a variety of factors, including rising global temperatures, pollution, overexploitation of resources, and the spread of diseases. This ongoing trend poses a significant threat to the survival of many mammal species, potentially leading to their extinction and disrupting ecosystems that rely on these species. Addressing this issue effectively requires robust methods for identifying and monitoring mammal species to facilitate conservation efforts. In this context, automatic species identification through advanced deep learning techniques offers a promising solution. This study explores the development and evaluation of three deep neural network models—VGG-19, ResNet-50, and Xception for the automatic classification of mammal species. These models were trained using a comprehensive benchmark dataset sourced from Kaggle, which includes images labeled with 45 different mammal species. The performance of the models was assessed based on their classification accuracy, with VGG-19 achieving 72%, ResNet-50 reaching 84%, and Xception attaining the highest accuracy of 93%. Notably, Xception, a depthwise convolution neural network, outperformed the other models, demonstrating superior effectiveness in classifying mammal species compared to the pointwise convolution neural networks used in VGG-19 and ResNet-50. This study underscores the potential of advanced deep learning techniques in enhancing species identification and conservation efforts.