CAT BREED AND EMOTION DETECTION USING YOLOv8 AND CNN

Abstract

This project presents a deep learning-based system designed to automatically detect both the breed and emotional state of cats from images. The proposed system integrates YOLOv8 (You Only Look Once), a model for identifying the presence of a cat in an image and classifying its breed, along with Convolutional Neural Networks (CNN) for recognizing the emotional state based on facial features. Initially, the input image undergoes preprocessing steps such as resizing and normalization to ensure consistent input for the models. The YOLOv8 model then detects the cat, generates a bounding box around it, and predicts its breed using learned visual characteristics. The detected region is extracted and passed as input to the CNN model, which analyzes facial patterns and expressions to classify emotions such as happy, sad, and angry. By focusing only on the relevant cat region, the system effectively reduces background noise and enhances prediction accuracy. The integration of object detection and deep learning-based classification enables the system to learn complex visual features and improve overall performance. Experimental results demonstrate that the proposed approach provides accurate and efficient predictions for both breed and emotion detection. This system highlights the potential of artificial intelligence in understanding animal behavior and can be further extended for real-time applications in pet monitoring, veterinary diagnosis, and animal welfare analysis.