🟢 Surface Defect Classification with CNN

This project uses a Convolutional Neural Network (CNN) to classify industrial surface defects in steel images.

✅ Dataset: NEU Surface Defect Database
Classes include:

• Crazing
• Inclusion
• Patches
• Pitted Surface
• Rolled-in Scale
• Scratches

✅ How to Run:

1. Download the NEU dataset and place it in your project folder.
2. Install dependencies:
3. Run the training script or notebook:
4. The trained model will be saved as surface_defect_model.h5.

✅ Model Summary:

• Input size: 150x150 RGB images
• 2 Convolutional + MaxPooling layers
• Dense layers with dropout
• Softmax output over 6 defect classes

✅ Result: Achieves good classification accuracy on validation data.

---

Example prediction:

---

Feel free to use or improve this project!