The growing integration of companion animals, such as dogs and cats, into households and society, alongside increasing policies and laws for pet management, necessitates reliable identification systems. Traditional methods, such as collars, microchips, and tattoos, pose issues such as invasiveness, the need for specialized scanners, and the potential for loss or error. This study introduces an AI-powered biometric identification system using smartphone images to offer a convenient and accessible alternative. We present a unified framework incorporating two advanced models: Dog Nose Network Version 2 (DNNetV2) for dog nose-print recognition and Cat Face Network (CatFaceNet) for cat facial identification. DNNetV2, the second iteration of our dog nose identification model, enhanced by the Tiny Vision Transformer and a novel MagFace-based loss function, achieves 99.8% Rank-1 identification accuracy and 99.62% verification accuracy. Similarly, CatFaceNet, leveraging human facial recognition techniques and our advancements in DNNetV2, achieves 99.96% Rank-1 identification accuracy and 99.26% verification accuracy. This comprehensive system demonstrates robust, high-performance identification for both dogs and cats, with potential applications in national pet registry, pet insurance, veterinary care, and lost pet recovery. Our unified solution establishes a new benchmark in pet biometric identification, ensuring exceptional reliability and performance.

Image classification is a fundamental task in computer vision, with convolutional neural networks (CNNs) being the state-of-the-art approach for this task. However, CNNs can be large and computationally expensive, which limits their deployment on resource-constrained devices. To address this issue, model optimization and compression techniques such as knowledge distillation and channel attention have been proposed. In this paper, we propose a modified version of the skip-connection knowledge distillation method, which integrates the efficient channel attention module. Our experiments on the CIFAR-10 dataset show that this approach improves the performance of both large teacher-small student and self-distillation scenarios, and also leads to smaller model sizes. We also visualize the intermediate feature maps to understand the learning process of the network. Our results demonstrate the effectiveness of combining knowledge distillation and channel attention for improving the performance and efficiency of CNNs.