Defect and non-defect image classification system of C-SAM images using artificial intelligence techniques

Bala Krishnan, Divien Raj Kumar (2024) Defect and non-defect image classification system of C-SAM images using artificial intelligence techniques. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Defect and non-defect image classification system of C-SAM images using artificial intelligence techniques.pdf - Submitted Version Download (2MB)

Abstract

This research presents an innovative method for automating the classification system in the semiconductor industry, addressing the inefficiencies and high costs associated with manual inspection of C-SAM images. The developed image classification system leverages modern technologies such as Artificial Intelligence (AI) for object detection and TensorFlow for data recognition and pre-processing. Its primary objective is to automate the identification, determination, and classification of C-SAM images into categories of defects and non-defects. The system enhances data gathering, annotation, model selection, training, and software configuration on a Convolutional Neural Network (CNN) architecture through thorough data pre-processing and customization. This study highlights the inefficiencies of conventional neural network techniques and the adverse effects of manual inspection, emphasizing the impact on time and cost. By automating the image classification process, the project aims to reduce the need for human labor, improve operational efficiency, and provide a user-friendly solution. The scope of the study includes offering employees’ instructions on data labeling and customization, restricting the model training learning rate, and specifying the number of epochs implemented. The project's objectives are to develop an algorithm capable of differentiating between defect and non-defect images using AI techniques, evaluate the classification performance of the developed algorithm, and provide a potential solution to improve the employee experience in the semiconductor sector. Additionally, the project includes implementing the system on a Graphical User Interface (GUI) to enhance user interaction and usability.

Actions (login required)

View Item