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
28 lines
923 B
Markdown
28 lines
923 B
Markdown
# Known Limitations at M3
|
|
|
|
These limitations are intentional and must not be treated as bugs during the first build.
|
|
|
|
## Real-time global data ingestion is out of scope
|
|
|
|
V1 focuses on selected Kempen areas.
|
|
|
|
## LiDAR is out of scope for V1
|
|
|
|
The architecture reserves space for PDAL/laspy, but V1 should not implement LAS/LAZ processing.
|
|
|
|
## Full model training is out of scope for V1
|
|
|
|
V1 may include model registry and inference interfaces. Training Studio is a later phase.
|
|
|
|
## Sentinel is not V1-critical
|
|
|
|
Remote sensing indices are valuable, but the first recruiter-focused demo should prioritize building detection, GRB validation and QA/QC.
|
|
|
|
## Fixture detector is allowed only as a bridge
|
|
|
|
A deterministic fixture detector is acceptable to validate the pipeline before YOLO is wired in. It must not be marketed as AI output.
|
|
|
|
## Reports are secondary
|
|
|
|
Exports are required. Rich PDF reporting is a later enhancement.
|