AI Vision Research & Evaluation

Identification Methodology & Scoring

How BrickFinder computes match confidence scores, extracts stud matrix vectors, evaluates color histograms, and handles edge curvature analysis.

1

Bounding & Segmentation

Isolates the brick contour from background noise, shadows, and surface glare using adaptive thresholding.

2

Stud Matrix Extraction

Identifies circular stud patterns, stud pitch ratios, and top/side/bottom spatial orientation.

3

3D Geometric Matching

Compares aspect ratios, bevel angles, anti-stud tubes, and slope degrees against LDraw 3D CAD meshes.

4

Confidence Scoring

Generates a normalized confidence percentage (e.g. 96%) and ranks candidate matches.

Confidence Score Interpretation

What match scores mean when reviewing identification results in BrickFinder.

90% – 99% Match Confidence

Definite Match

Geometric contour, stud matrix, and color swatch align perfectly with high CAD correspondence.

96%
75% – 89% Match Confidence

Likely Candidate

Shape and category match, but secondary features (such as printed sub-variants or color shifts) require user review.

82%
Below 75% Match

Stumped / Partial Match

Shadows, angle obscuration, or custom MOC modifications present. Use our Reverse Underside Finder or Part Lookup tool.

<75%