Cheetah Face ID

Our AI-driven process for identifying cheetahs starts with a segmentation model using transfer learning from Roboflow 3.0 Instance Segmentation, achieving a high accuracy of 90.3% mAP to precisely extract cheetah faces from images. We then employ a Siamese neural network with triplet loss to create detailed facial embeddings for each cheetah. For identifying individual cheetahs in new images, we use the K-nearest neighbor algorithm on these embeddings, ensuring accurate recognition. Experience our technology's capabilities by trying out our demonstration model, trained on five distinct cheetahs, showcasing our AI's precision in wildlife identification