Crack Analyzer

This project provides a Gradio-based interface for inspecting binary crack images (white = crack, black = background). It extracts a skeleton of the crack network, finds the maximum crack width, detects branch/end points, computes geometric features, and checks compliance against user defined limits.

Features

• Skeleton Extraction using skimage.morphology.thin.
• Maximum Crack Width located with a KD-tree and visualized on the image.
• Branch and Endpoint Detection via a 3×3 convolution overlaying green (end) and red (branch) dots.
• Geometric Metrics including area, length, average width, maximum width, endpoint count, branch point count and an estimated branch count.
• Compliance Checking against limits for maximum/average width, area ratio and crack length.

Running

Install dependencies:

pip install -r requirements.txt

Start the app:

python app.py

Upload a binary crack image and adjust the pixel size and compliance thresholds as needed. The app will display the original image, the extracted skeleton, branch/endpoint overlay and maximum width visualization along with computed metrics.