Smart Fruit: An AI-Powered System for Real Time Fruit Classification and Detection

SmartFruit is an AI-Powered System for Real- Time Fruit Classification and Detection," aims to develop a highly efficient and scalable solution for automatically identifying and categorizing different types of fruits using deep learning techniques. Leveraging the capabilities of an ESP32 module for local image capture and AWS cloud services for processing, the system is designed to perform real-time fruit classification and detection with high accuracy. Two datasets are utilized: a public dataset (FIDS:30) with 30 fruit class and a customized datasets with 6 fruit category, enhanced through data augmentation techniques to improve model robustness. The project employs advanced deep learning models, including YOLOv3, YOLOv7 for object detection, and ResNet50 and VGG16 for image classification. The deployment strategy involves using the ESP32 module for capturing images and performing preliminary data processing, followed by transmitting data to AWS for complex model inference using AWS Lambda and AWS SageMaker. This hybrid approach combines the low-power advantages of edge computing with the scalability and processing power of cloud services. The system's realtime capability and high accuracy make it suitable for diverse applications in agriculture, supermarkets, and educational tools. By deploying deep learning models in a distributed architecture, this project highlights a cost-effective and innovative solution for fruit classification and detection, paving the way for further advancements in smart agricultural practices and automated retail management systems. The matter presented in this article is the final year project work done by the UG students.