AI-powered breed identification and personalized care

Wong, Kitty Ai Lin (2025) AI-powered breed identification and personalized care. Final Year Project, UTAR.

PDF Download (7Mb)

Abstract

This project investigates the integration of deep learning and modern artificial intelligence to enhance mobile pet care applications. The proposed system employs the MobileNetV2 architecture, a lightweight and efficient Convolutional Neural Network(CNNs) , to automatically identify pet breeds from photographs captured or uploaded by users. To provide foundational breed information, the application incorporates data from Dog API by Kinduff, Dog CEO and API Ninjas Dogs, which offer structured trait ratings such as shedding levels and energy characteristics along with general breed care tips. These sources supply only broad, breed-specific guidance and are implemented as fixed content within the application. To deliver care recommendations that go beyond generic advice, the system integrates the Google Gemini large language model, allowing users to request personalised guidance that considers their own pet’s breed, age and medical records. This dual approach addresses key limitations of existing pet care applications, which often rely on manual breed entry and offer only general advice, by combining automated breed recognition with customised, context-aware care recommendations. As a result, the system reduces human error, improves the reliability and relevance of pet health recommendations and simplifies overall pet management, thereby promoting better pet well-being and enhancing user satisfaction.

Actions (login required)