AI PET EMOTION RECOGNITION SYSTEM USING DEEP LEARNING

Understanding pet emotions is crucial for enhancing animal welfare and strengthening human-pet bonds, yet objective emotion assessment remains challenging. This project addresses this need by developing a comprehensive pet emotion detection system using real-time TensorFlow Lite inference on mobile devices. The system employs a Convolutional Neural Network trained on dog facial expressions to classify emotions into four categories: angry, happy, relaxed, and sad. The implementation addresses critical mobile machine learning challenges, including preprocessing pipeline alignment, model optimization, and real-time inference performance. Through systematic debugging and Python-like preprocessing implementation, the system successfully eliminated model bias and achieved accurate emotion classification, providing pet owners with valuable insights into their pets’ emotional well-being.