SmartPaw assistant: intelligent pet care through image recognition and machine learning

Lim, Xin Yi (2026) SmartPaw assistant: intelligent pet care through image recognition and machine learning. Final Year Project, UTAR.

PDF Download (8Mb)

Abstract

The surge in pet ownership has revealed some constraints in pet care applications, such as overdependence on manual data entry, inadequate personalisation and lack of tools for understanding pet sentiment. The contribution of this work is the SmartPaw Assistant, an AI-based mobile application using Flutter, for the integrated solution. The app offers several functionalities, such as analysing pet food nutrition with image recognition, evaluating health status through image scanning, generating automated timeline-based health reports, calendar reminders and an expense tracker, as well as a community interaction and a price comparison tool. At the heart of the framework lies a robust dual-model pet emotion interpretation functionality, which is a CNN-BiLSTM model for audio classification that reaches an accuracy of 89% and a MobileNetV2 model for visual body language recognition that reaches an accuracy of 87%. To improve user-specific adaptability, a personalisation strategy based on transfer learning is exploited for offline model updates by utilizing user feedback every two weeks. This process enables an increasing and near-perfect accuracy for each pet. The backend is built with FastAPI and Supabase Database with Ngrok connecting during development. In general, SmartPaw Assistant showcases the promise of an AI-powered and comprehensive pet care service that minimizes manual control, personalises and improves the effectiveness of pet care management.

Actions (login required)