Use class schema CSV/JSON to map vector strings or raster pixel values to target class IDs. Recommended explicit mapping columns are source_value and target_class_id. Legacy aliases such as merge_to are still accepted, but new schemas should use target/new class IDs.
By default, 0 is treated as background/no label in hard-label rasters. Enable “Treat 0 as class” when 0 is a real class. For example, binary 0/1 can be remapped to 1/2 or kept as 0/1 depending on your modelling convention.
| Layer type | NoData convention | Why |
|---|---|---|
| Hard class-ID raster | Background/no-label, usually 0 | 0 is normally used as no label. |
| One-hot Byte stack | 255 | 0 is valid absence of a class. |
| Probability/support/context/weight/helical Float32 | -9999 | 0 is a valid probability, risk, uncertainty, support or feature value. |
| Edge-risk Byte raster | 255 | 0 is valid “no edge risk”. |
Inputs should have matching CRS and compatible grid alignment where possible. HELIX can align rasters, but pre-aligned categorical rasters reduce resampling artefacts. Preflight reports CRS mismatch, non-overlap, pixel-size differences and origin misalignment.
Hauser, S., Augner, L., & Schmitt, A. (2025). Perfect Labelling: A Review and Outlook of Label Optimization Techniques in Dynamic Earth Observation. Remote Sensing, 17(7), 1246. doi: 10.3390/rs17071246.
Hauser, S. (2025). Automated Feature & Label Refinement in the Context of Environmental Monitoring by Multi-Modal Satellite Data. Dissertation, Karlsruher Institut für Technologie. doi: 10.5445/IR/1000185980.
Hauser, S., Dachsberger, S., Schmitt, A., & Hinz, S. (2026). Calibrated U-Net with HELIX-Based Label Enrichment for Ageing-Aware Spatio-Temporal Urban Change Detection. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Accepted / in press; DOI not publicly available yet.