ajandClaude Opus 5.5 0be9e2201a feat(training): add sketch-based drawing search prototype ...

Python tool that finds drawings resembling a hand sketch, photo or MS Paint
outline. It indexes the OpenNest server's drawing library (read-only) and/or
DXF folders, compares DINOv2 embeddings of normalized renderings across all
8 rotations/mirrors, and writes an HTML page of the closest matches with an
optional overall-size filter.

train.py fine-tunes the embedding contrastively on synthetic sketches of the
library itself (no labelling); evaluate.py measures recall on held-out
drawings or a folder of real sketches. On 40,757 server drawings, held-out
synthetic sketches find the right shape in the top 20 75% of the time,
up from 18% zero-shot.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

2026-10-28 06:09:40 -04:00

..

Data

style: apply CSharpier formatting to all C# sources

2026-09-20 16:41:50 -04:00

notebooks

feat(training): add training notebook skeleton and requirements

2026-03-14 20:27:41 -04:00

sketch-search

feat(training): add sketch-based drawing search prototype

2026-10-28 06:09:40 -04:00

OpenNest.Training.csproj

refactor: extract training data collection into OpenNest.Training

2026-03-14 14:41:38 -04:00

Program.cs

refactor(io): persist best fits through an explicit IBestFitStore

2026-10-16 14:54:43 -04:00

TrainingDatabase.cs

refactor(engine): align namespaces with directory layout under OpenNest.Engine

2026-09-21 13:18:34 -04:00