CrackScan - Intelligent Crack Detection System
<div align="center">
Deep Learning-Based High-Precision Crack Detection System
Lightweight Model · High Accuracy · Image Enhancement · Multi-Dataset Validation
中文 | English
</div>
🌟 Highlights
- ✅ Lightweight Model: Only 45.2 MB, 11.8M parameters
- ✅ High Accuracy: mIoU 85.4%, outperforms traditional methods
- ✅ Image Enhancement: Retinex algorithm, 15-20% improvement in low-light scenarios
- ✅ Multi-Dataset Validation: Validated on concrete, metal, wood, and other materials
- ✅ Fast Inference: < 2 seconds per image (GPU)
- ✅ Batch Processing: Automatic multi-folder processing
📊 Performance Metrics
| Metric | Value |
|--------|-------|
| mIoU | 85.4% |
| Precision | 92.3% |
| Recall | 89.7% |
| F1-Score | 90.9% |
| Model Size | 45.2 MB |
| Inference Speed | 1.8 sec/image |
📸 Detection Results
Concrete Cracks
<table>
<tr>
<td align="center"><b>Original</b></td>
<td align="center"><b>Enhance</b></td>
<td align="center"><b>Prediction</b></td>
<td align="center"><b>Visualization</b></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_cracks/00014_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00014_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00014_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00014_visualization.jpg" width="250"/></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_cracks/00015_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00015_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00015_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00015_visualization.jpg" width="250"/></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_cracks/00016_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00016_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00016_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_cracks/00016_visualization.jpg" width="250"/></td>
</tr>
</table>
<table>
<tr>
<td align="center"><b>Original</b></td>
<td align="center"><b>Enhance</b></td>
<td align="center"><b>Prediction</b></td>
<td align="center"><b>Visualization</b></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_iron/3d6fc6cb2_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d6fc6cb2_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d6fc6cb2_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d6fc6cb2_visualization.jpg" width="250"/></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_iron/3d7d0609f_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d7d0609f_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d7d0609f_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d7d0609f_visualization.jpg" width="250"/></td>
</tr>
<td><img src="data/prediction_results/kaggle_iron/3d8f3ef8c_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d8f3ef8c_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d8f3ef8c_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_iron/3d8f3ef8c_visualization.jpg" width="250"/></td>
</tr>
</table>
<table>
<tr>
<td align="center"><b>Original</b></td>
<td align="center"><b>Enhanced</b></td>
<td align="center"><b>Prediction</b></td>
<td align="center"><b>Visualization</b></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_wood/99900054_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900054_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900054_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900054_visualization.jpg" width="250"/></td>
</tr>
<tr>
<td><img src="data/prediction_results/kaggle_wood/99900055_original.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900055_enhanced.jpg" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900055_prediction.png" width="250"/></td>
<td><img src="data/prediction_results/kaggle_wood/99900055_visualization.jpg" width="250"/></td>
</tr>
</table>
<div align="center">
</div>