Classification of Endangered Species of Wild Cats through Xception Transfer Learning Model

In recent years, populations of wild cats such as Caracals and African Leopards have dropped low enough that many wild cats have been labelled as endangered or critically endangered by the International Union of Conservation of Nature's Red List of Endangered Species. Caused due to direct or indirect human activities such as hunting and poaching, and industrialization leading to habitat destruction, it has become vital to monitor and manage these majestic creatures for their conservation. Identifying and classifying wild cats from images is a difficult feat that demands reliable and effective deep learning (DL) techniques. This work demonstrated a method for identifying and categorizing wild cats using 1439 photos with 224×224×3 pixel resolution and ten annotated classes, including African Leopard, Caracal, Cheetah, Clouded Leopard, Jaguar, Lion, Ocelot, Puma, Snow Leopard, and Tiger. The Xception transfer learning (TL) model, a Convolutional Neural Network (CNN) that has been pre-trained and has demonstrated outstanding performance on several image categorization tasks was employed. Using the TL strategy, the Xception model was optimized on the dataset of wild cat images. Data augmentation methods were also used to broaden the variety of the training dataset and enhance the model's generalizability, such as rotation, zooming, and horizontal flipping. The test findings demonstrate that the suggested approach successfully detects and classifies wild cats from the images with an accuracy of 96%, a sensitivity of 96%, and a specificity of 99.4%. The proposed methodology could be implemented to identify and monitor various wild cat species in their natural habitats for wildlife conservation, monitoring, and management.