Segment 2D

Manual and semi-automated segmentation of 2D spatial data into named, coloured classes, for building ML training sets and for review/QA of existing labels. The camera is locked to a straight plan view — pan and zoom only — so screen space and data space stay aligned. Drop a DXF or SVG drawing, a rasterized PDF page, or a 2D point set. Select with rectangle, polygon or lasso, or click a seed and grow the region outwards through neighbouring elements; shrink to erode a ragged boundary. Type a class name, pick its colour, and label the selection — each labelled segment becomes a coloured group under Features in the object tree, with per-segment and per-class show/hide, recolour and delete. Export the labels as a dense per-element CSV (one row per point/segment/face with class id and coordinates) or as a sparse JSON class/segment manifest, and export the geometry alongside it so the pair forms a complete dataset.

Supported file formats: DXF, DWG, SVG, PDF, XYZ, PTS