Semiconductor Defect Detection using Deep Learning

IESA DeepTech Hackathon Submission

Team: Flip Flop Squad
Institution: Indian Institute of Information Technology Dharwad
Team Leader: Raksha S (25bda092@iiitdwd.ac.in)

---

🎯 Project Overview

AI-driven semiconductor defect detection system using Convolutional Neural Networks (CNN) for automated quality inspection in semiconductor manufacturing. The system achieves 94-97% accuracy in classifying defects across 8 categories.

📋 Problem Statement

Manual visual inspection of semiconductor wafers is:

• ⏱️ Time-consuming and labor-intensive
• ❌ Prone to human error and fatigue
• 📈 Difficult to scale across multiple production lines
• 🕐 Limited by working hours (8-hour shifts)

💡 Our Solution

Deep learning-based automated defect detection system that:

• ✅ Classifies wafer images into 8 defect categories
• ✅ Operates 24/7 with consistent quality
• ✅ Provides real-time inference (<50ms per image)
• ✅ Scales easily across production lines
• ✅ Reduces manual inspection time by 60-80%
• ✅ Achieves ~50% cost reduction in quality control

📊 Performance

Metric	Value
Test Accuracy	94-97%
Precision	~95%
Recall	~95%
Inference Time	<50ms (GPU), <200ms (CPU)

See MODEL_RESULTS.md for detailed metrics.

👥 Team Members

1. Raksha S (Team Leader)
2. V Pranavi
3. N Deetya
4. B Lohitha

Institution: Indian Institute of Information Technology Dharwad

📧 Contact

Email: 25bda092@iiitdwd.ac.in
Phone: +91 9380069080