Cat Body Language Recognition Using Computer Vision in an Android Application

Understanding cat behaviour is essential for fostering healthy human-cat relationships, but its inherent complexity frequently leads to misunderstandings. This study introduces Emeowtions, an innovative Android application employing artificial intelligence (AI) to decipher cat emotions and body language in real-time. Addressing market gaps for comprehensive tools, Emeowtions integrates the YOLOv8n object detection model with a custom-trained multi-label classification model for cat emotion and body language analysis. The custom model was developed based on the CRoss Industry Standard for Data Mining (CRISP-DM) framework and trained using transfer learning with MobileNetV3 on a custom curated dataset of annotated cat images. Built using the Waterfall methodology, the application allows users to obtain real-time, AI-driven insights via their smartphone camera. Beyond that, it provides a hybrid recommendation system suggesting tailored behaviour suggestions, a user feedback loop for model refinement, and a direct chat interface for veterinary consultations. Technical evaluation showed the AI model achieved a recall of 0.742. Overall, Emeowtions offers a valuable, practical tool that demonstrates AI's capability to reduce misinterpretations of cat behaviour, ultimately fostering healthier human-animal relationships and contributing to improved cat welfare.