Skip to content

post-inference.md

Latest commit

 

History

History
73 lines (41 loc) · 2.35 KB

post-inference.md

File metadata and controls

73 lines (41 loc) · 2.35 KB

Working in progress

Post inference processing

Vectorization:

Feedback: rasterio.features.shapes or gdal_polygonize produces "stair-step" polygons. Apply a simplification algorithm (e.g., Douglas-Peucker) during vectorization to reduce file size and make the map look cleaner.

Merging strategy: Maximum

For pixels covered by multiple tiles, take the maximum probability:

P_merged(x, y) = max(P_tile1(x, y), P_tile2(x, y), ...)

Rationale: Consistent with the detection philosophy—if any tile view detects RTS, include it.

Scale Fusion

Combine predictions across scales using pixel-wise maximum:

P_final = max(P_1.0, P_0.5)

Rationale: If any scale confidently detects RTS, include it. Maximum operation is conservative toward detection while individual scale thresholds control precision.

Prediction Merging

Overlap Handling

Adjacent (4-neighbours) tiles overlap by 50%. The overlapping regions have multiple predictions that must be merged.

  • can adjust to 25% overlap if computation resource is limited (estimate total tile numbers and gpu-hour for both 25% and 50%)

Merging Procedure

Option A: On-the-fly merging (memory-efficient)

  1. Create output raster for region with NoData fill
  2. For each tile prediction:
    • Read overlapping region from output
    • Compute pixel-wise maximum with new prediction
    • Write merged result back
  3. Advantage: Low memory; Disadvantage: Many I/O operations

Option B: Batch merging (faster)

  1. Accumulate all tile predictions in memory (or memory-mapped file)
  2. Apply reduction (maximum) across overlapping tiles
  3. Write final merged raster
  4. Advantage: Faster; Disadvantage: High memory for large regions

Recommendation: Use Option B for manageable regions (e.g., per Arctic subregion), Option A for full pan-arctic if memory-constrained.

Area and Perimeter Calculation

EPSG:3857 distorts areas at high latitudes (~13x inflation at 74°N). All area and perimeter measurements must use geodesic calculations (e.g., pyproj.Geod.geometry_area_perimeter) or reproject to a local equal-area CRS — never compute directly from EPSG:3857 coordinates.

Output Chunking

For pan-arctic scale, produce merged outputs per region rather than single global raster:

  • Easier to manage and distribute
  • Enables parallel processing
  • Allows region-specific quality control