wafer-defect-detection

Mask-aware wafer defect classification using a DenseNet-based CNN with explicit geometry masking and Grad-CAM explainability.

Enterprise-grade ML platform for semiconductor wafer defect classification using ResNet-50 U-Net architecture with active learning.

Drift-Sense solves the Navigation-Error Recovery problem. By utilizing a hybrid architecture of classical vision (NCC) and a lightweight Siamese CNN ranker, it accurately matches high-resolution reference patterns against noisy, low-resolution SEM images without heavy transformers.

GPU-accelerated computer vision for semiconductor wafer defect detection using YOLOv8-L, TensorRT, Triton, and MLOps

🚀 Official code for IEEE ECCE 2025 paper: "Advancing Semiconductor Fabrication: A CNN-Based Wafer Defect Detection with XAI Insights". ResNet50 classifier (94.08% accuracy) with Explainable AI (Grad-CAM) on the WM-811K benchmark.

晶圆 SEM 缺陷分析桌面演示：分类、语义分割、形貌描述与检索流程；不含保密数据和检索图库。

ResNet-18 wafer defect classifier on WM-811K: focal loss, CBAM attention, temperature-scaled calibration, Grad-CAM++. Macro-F1 0.92

Wafer Guard is a machine-learning-based semiconductor wafer defect detection system that analyzes sensor data to identify wafers requiring further inspection.

SimCLR self-supervised pretraining on 638k unlabeled WM-811K wafer maps, plus multi-seed ensembling, extending wafer-defect-classifier

CNN-based semiconductor wafer defect classification with 91.57% test accuracy and Streamlit deployment.

Research-driven wafer defect classification framework combining classical ML and CNN-based approaches, with experimental analysis of few-shot learning under severe class imbalance and limited data regimes.