Jens
faeb58ef6d
GeoIntel release gates / Compile, test, contracts and builds (push) Successful in 1m49s
GeoIntel release gates / Python and npm vulnerability policy (push) Successful in 21s
GeoIntel release gates / Production AI image, SBOM and container scan (push) Successful in 5m39s
GeoIntel release gates / Deploy exact gated revision to Unraid (push) Failing after 58m43s
38 lines
738 B
Markdown
38 lines
738 B
Markdown
# Detection Classes Catalog
|
|
|
|
## V1 required class
|
|
|
|
### building
|
|
|
|
Reason:
|
|
Buildings are the best first detection target because GRB can be used as a reference layer for QA/QC.
|
|
|
|
Outputs:
|
|
|
|
- bounding box polygon
|
|
- confidence
|
|
- source tile
|
|
- model id
|
|
|
|
## V1 optional classes
|
|
|
|
### solar_panel
|
|
|
|
High portfolio value but dataset/model availability may be harder.
|
|
|
|
### vehicle
|
|
|
|
Useful for generic object detection demo but less central to GIS validation.
|
|
|
|
### tree
|
|
|
|
Potentially useful but segmentation may be more appropriate than bounding boxes.
|
|
|
|
## Not required for V1
|
|
|
|
- roads as object detection class
|
|
- water as object detection class
|
|
- agriculture as object detection class
|
|
|
|
These are better handled through segmentation or raster indices.
|