Development of low-cost image classification system using ESP32-CAM

Kamaruddin, Muhammad Adni (2024) Development of low-cost image classification system using ESP32-CAM. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full text) Development of low-cost image classification system using ESP32-CAM.pdf - Submitted Version Download (2MB)

Abstract

Advances in Artificial Intelligent (AI) and the availability of large training data set shave made image classification popular in various domains. The market for AI in enterprise applications is growing rapidly, indicating the potential for further development in image classification field. Nevertheless, the high cost in developing an image classification system could hinders its widespread adoption and becomes more challenging to support Industrial Revolution (IR) 4.0. Therefore, in this project, we proposed a low-cost image classification system using ESP32-CAM which capable of performing real-time image classification based on the trained model. The method involves training a deep learning model (i.e., microneural network) using a dataset of non-defective/defective integrated circuits (ICs) on Edge Impulse platform. Subsequently, deploying the generated code of the successful trained model on ESP32-CAM. The model is optimized to fit the limited resources (i.e., memory and processing power) of the ESP32-CAM. By using the built-in camera in ESP32-CAM, it can performs real-time image classification of non-defective/defective ICs. The proposed system achieves 86.1% prediction accuracy by using 1571 image data set of defective and non-defective ICs.

Actions (login required)

View Item