Skip to contents

segmantR (development version)

Bug fixes

  • sg_segment_threshold(method = "adaptive") no longer errors with “replacement has length zero”. The internal integral-image lookup now treats out-of-range indices as contributing zero, so adaptive thresholding works for all images.

segmantR 0.1.0

  • Initial release.
  • Classical segmentation: adaptive thresholding, watershed, Voronoi propagation.
  • Deep learning wrappers: Cellpose, StarDist, Mesmer (via reticulate).
  • Human-in-the-loop annotation and correction workflow.
  • Custom model training and portable model export (.segmantR archives).
  • Feature extraction: intensity, morphology, texture, location.
  • Shiny application with six tabs for interactive segmentation workflows.
  • Export to GeoJSON, QuPath, CSV, SpatialExperiment.
  • Bundled synthetic sample data for demonstrations.