How ayushgigras/crackvision-ai works
CrackGauge is a command-line computer-vision tool for turning concrete images into crack masks and width, length, and orientation measurements, optionally scaled to millimetres. The principal workflow begins at main.py, which orchestrates preprocessing, segmentation, measurement, and calibration. The README also describes PDF inspection reports, but no report module appears in the supplied tree or excerpts; that capability is shown as documented rather than verified implementation. Edges reflect the shown orchestration and documented calibration options; unsampled behavior is not inferred.
Main entry points: main.py Pipeline runner · preprocessing.py Image preprocessing · segmentation.py Crack segmentation · measurement.py Width, length, orientation
Components 7
Pipeline entry
- Pipeline runner
main.py - PDF inspection report
main.py
Image analysis
- Image preprocessing
src/preprocessing.py - Crack segmentation
src/segmentation.py
Crack metrics
- Width, length, orientation
src/measurement.py
Physical calibration
- Scale and ArUco calibration
src/calibration.py
Other components
- Inspector
Connections 6
- Inspector → Pipeline runner: selects image
README.md - Pipeline runner → Image preprocessing: runs preprocessing
main.py - Pipeline runner → Crack segmentation: runs segmentation
main.py - Pipeline runner → Width, length, orientation: runs measurement
main.py - Pipeline runner → Scale and ArUco calibration: resolves scale
main.py - Pipeline runner → PDF inspection report: generates report
README.md