Pipeline Execution

Running Custom Trained Neural Network
Executing Advanced Post-Processing
Calculating Geometric Metrics
Estimated time: ~30 seconds

Stone Vein Segmentation & Geometric Analysis

GO TO LIVE DEMO
Sample Raw Input Original Texture
AI Pipeline
  • Training: Model pre-trained on the Kaggle Crack Segmentation dataset.
  • Inference: The model trained on cracks, applied to detect stone veins with little labeled data.
  • Post-Processing: Refined output using advanced Image Processing techniques.
  • Validation: Performance metrics calculated against a Custom Ground Truth mask.
  • Computer Vision Algorithms:
    View FullSource on GitHub & Model Training Configurations
Sample Summary Analysis
1. Model Prediction
Test Image F1-Score (Dice)
0.710
2. Geometric Analysis
TP
FN
FP
3. Topological Skeleton
Live Validation Sandbox

STEP 1: Upload your stone image.

STEP 2 (Validation): To provide a Ground Truth, you can DRAW directly on the image below OR UPLOAD a precise mask file.