RostiS/pavement_cell_segmentation

🛣️ Pavement Cell Segmentation

This Hugging Face Space provides an interactive demo for segmenting pavement cells in road surface images using a deep learning model. The model uses a U-Net architecture converted to ONNX format for efficient inference.

Features

• —Image Upload: Upload pavement images in various formats (JPG, PNG, TIFF, BMP)
• —Parameter Tuning: Adjust processing parameters for optimal results
• —Real-time Processing: Get segmentation results with visual feedback
• —Download Results: Download the segmentation mask as a PNG image
• —Overlay Visualization: View the segmentation overlay on the original image

How to Use

1. 1.Upload Image: Click on the upload area and select a pavement image
2. 2.Adjust Parameters:
3. 3.Tile Size: Size of image tiles for processing (128-512 pixels)
4. 4.Tile Overlap: Overlap between adjacent tiles (32-128 pixels)
5. 5.Threshold: Segmentation threshold for binary mask (0.1-1.0)
6. 6.Show Overlay: Toggle overlay visualization
7. 7.Process: Click "Process Image" to run the segmentation
8. 8.Download: Save the segmentation mask using the download button

Parameters Guide

Tile Size

• —Large tiles (512px): Best for high-resolution images, slower processing
• —Medium tiles (256px): Good balance for most images (recommended)
• —Small tiles (128px): Faster processing, may miss larger patterns

Tile Overlap

• —High overlap (64-128px): Better accuracy at tile boundaries, slower processing
• —Medium overlap (64px): Good balance (recommended)
• —Low overlap (32px): Faster processing, may have boundary artifacts

Threshold

• —High threshold (0.8-0.9): Cleaner results, may miss subtle features
• —Medium threshold (0.6-0.8): Good balance (recommended)
• —Low threshold (0.3-0.6): More sensitive detection, may include noise