Wafer Defect Classification Using Deep Learning

A CNN-based wafer defect classification system developed using the WM-811K semiconductor wafer dataset.

Features

• 9 wafer defect categories
• Convolutional Neural Network (CNN)
• Streamlit Web Application
• Real-time prediction
• Accuracy: 80%
• Macro F1 Score: 0.79

Technologies

• Python
• PyTorch
• NumPy
• Scikit-Learn
• Streamlit

Dataset

WM-811K Semiconductor Wafer Dataset

Classes

• Center
• Donut
• Edge-Loc
• Edge-Ring
• Loc
• Near-full
• Random
• Scratch
• None

Author

Aman Raj B.Tech Electronics and Communication Engineering NIT Calicut

streamlit run app.py