Pet Breed Classification and Diet Recommendation Using Deep Learning

Abstract

The rising popularity of domestic pets has generated a new demand for more personalized care systems, especially regarding diet and exercise recommendations based on breed. In this paper, we present a new Pet Care Recommendation System which uses deep learning and advanced image processing techniques to identify pet breeds from images and make breed appropriate recommendations. The model utilizes EfficientNetB5, which provides a high degree of efficiency with 95% classification performance. The accuracy is a result of the design architecture, as well as being trained on a dataset that encompasses a variety of animal breed sets. After breed recognition, the model provides categorized diet plans, which outline recommendations for morning, afternoon, and evening meals per pet breed, along with a recommended walking schedule that fits the individual needs and expectations of a particular breed based on exercise knowledge. The recommendations are made specifically to enhance and/or improve a pet’s health and overall well-being. Additionally, the model can be used within a Flask application for real-time user engagement and convenience for pet owners. Through an automated pet identification process and recommendations, the user no longer searches manually, and also saves time while providing scientifically driven findings to enhance a pet’s health status. The scalability and flexibility of the application are viable solutions to the demands of the growing pet care industry. Future consideration involves extending the model for multiple types of pets, expanding health monitoring capabilities, and using multiple language options to provide for a broader global client base.