Quickstart: create a segmentation layer

Quickstart: create a segmentation layer

This quickstart gets you from a source image to a reviewable segmentation layer. It uses the current SAM 3 workflow as the model example, but the RasterForge steps are the same project flow: import, AOI, render, segment, review, and export.

  1. Open, create, or install a project
    Projects keep imagery, renders, segmentation profiles, layers, review state, edits, and exports together. New users can install a sample case from the home page or Projects page to inspect a prepared workflow before importing their own imagery.
  2. Import imagery
    Open Import and upload GeoTIFF, PNG, or JPEG imagery. GeoTIFF is preferred when map placement and geospatial export matter.
  3. Confirm bounds
    If RasterForge cannot place the image on the map, add WGS84 bounds before selecting a map AOI.
  4. Select an AOI
    Draw or reset the area of interest. The AOI controls the part of the image rendered and sent through segmentation.
  5. Generate a render
    Choose a render profile, inspect the preview, tune bands or stretch if needed, then continue when the segmentation render is ready.
  6. Run segmentation
    In Segment, choose Find things for repeated targets, Select objects for point or box guided active-image work, or Auto masks for class-agnostic object candidates.
  7. Inspect the layer
    Use the layer controls to compare GeoJSON and Pixel mask views, update GeoJSON, create a stored Pixel mask when available, and mark a review master when the layer is the output you want.
  8. Review and export
    Use Review to choose master outputs, inspect image overviews and coverage, then export GIS files or handoff packages from Export.

Start with a visible, repeated target. Water, cloud edges, vegetation blocks, roofs, buildings, ponds, or field boundaries are easier to inspect than vague classes like "interesting areas".

What to learn next