af0dbe863a
[AZ-338] [AZ-283] C2 NetVLAD mandatory simple-baseline VprStrategy
...
NetVLAD is the C2 comparative baseline per the engine rule (every
production-default backbone ships with a simple-baseline alongside).
Runs on the C7 PyTorch FP16 runtime (NOT TRT) so a TRT engine compile
bug cannot simultaneously break NetVLAD AND UltraVPR.
Production changes:
- c2_vpr/net_vlad.py — NetVladStrategy + module-level create() factory.
Constructor wires InferenceRuntimeCut + DescriptorIndexCut +
NetVladBackbonePreprocessor + DescriptorNormaliser + FaissBridge.
embed_query pipeline: preprocess -> runtime.infer -> dual-stage
normalisation (intra-cluster THEN global L2) -> VprQuery.
retrieve_topk delegates one-line to FaissBridge.
- c2_vpr/_net_vlad_architecture.py — Arandjelovic et al. 2016 NetVLAD
layer over torchvision VGG16 features + optional Linear PCA
projection to descriptor_dim (default 4096; published Pittsburgh
reference uses K*D=64*512=32768 raw + Linear(32768, 4096) PCA).
- c2_vpr/_preprocessor_net_vlad.py — OpenCV-based image preprocessor:
decode -> centre-crop square -> resize (480, 480) -> ImageNet
normalisation -> FP16 NCHW. Calibration is not consumed (NetVLAD
is calibration-agnostic per published preprocessing chain).
- c2_vpr/inference_runtime_cut.py — NEW AZ-507 consumer-side cut
mirroring C7 InferenceRuntime; lets c2_vpr stay AZ-507-clean.
- c2_vpr/config.py — added netvlad_descriptor_dim: int = 4096 knob.
- helpers/descriptor_normaliser.py — added intra_cluster_normalise
(DescriptorNormaliser v1.0.0 -> v1.1.0; backward-compatible add).
- runtime_root/vpr_factory.py — added _register_strategy_architecture
helper that binds (MODEL_NAME, architecture_factory(descriptor_dim))
to C7's architecture registry before delegating to the strategy's
create() factory. Keeps the c7 import at L4, preserves AZ-507.
- fdr_client/records.py — registered vpr.embed_query,
vpr.backbone_error, vpr.preprocess_error record kinds.
Tests:
- tests/unit/c2_vpr/test_net_vlad.py — 31 tests covering all 11 ACs +
preprocessor contract + architecture factory + constructor
validation + FDR record emission.
- tests/unit/test_az283_descriptor_normaliser.py — +8 tests for the
new intra_cluster_normalise.
- tests/unit/test_az272_fdr_record_schema.py — +3 fixture payloads.
Full unit suite: 1608 passed / 80 env-skipped (+43 new tests).
Per-batch code review (batch_46_review.md): PASS_WITH_WARNINGS
(4 Low-severity hygiene findings; no Critical/High/Medium).
Architectural notes:
- The spec implied c2_vpr.net_vlad.create() registers the architecture
with C7. That violates AZ-507 (no cross-component imports). Resolved
by exposing MODEL_NAME + architecture_factory(descriptor_dim) on the
strategy module and having the composition root perform the C7 bind.
- C7 PyTorch runtime API names in the spec (forward, load_engine)
were outdated; aligned implementation with the live v1.0.0 Protocol
(infer, compile_engine + deserialize_engine). Spec hygiene flagged
in review F2.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-05-13 22:30:29 +03:00
3acc7f33dd
[AZ-270] [AZ-272] [AZ-279] [AZ-281] [AZ-283] Compose root + FDR schema + 3 Layer-1 helpers
...
AZ-270: composition root with strategy registry, tier-gated lookup,
topo-order construction, all-or-nothing teardown, StrategyNotLinkedError
payload.
AZ-272: orjson-backed FdrRecord serialise/parse with forward-compat for
unknown payload + top-level fields and canonical overrun-record shape.
AZ-279: pyproj-backed WGS84/ECEF/ENU + OSM slippy-map tile math with
WgsConversionError for shape/range/zoom guards.
AZ-281: strict EngineFilenameSchema build/parse/matches_host with
anchored regex + enum validation; round-trip identity by construction.
AZ-283: dtype-preserving (fp16/fp32) single + batch L2 normaliser with
zero-norm safety and descriptor_metric() source-of-truth.
pyproject.toml pins pyproj>=3.6 and orjson>=3.9 (named-backend deps per
the AZ-272 / AZ-279 contracts). New DTOs LatLonAlt + BoundingBox and
EngineCacheKey + HostCapabilities land in _types/ to back the helper
contracts.
203 unit tests pass (64 new). Review verdict: PASS_WITH_WARNINGS;
findings are perf-NFR deferrals + dep amendment + minor docstring polish.
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-05-11 02:03:36 +03:00
b12db61444
[AZ-263] Bootstrap: repo skeleton + Docker + CI + Alembic + Tier-1 tests
...
Implements the AZ-263 / E-BOOT initial structure task:
- Python src/-layout package `gps_denied_onboard/` with per-component
interface stubs (14 components), type-only DTOs under `_types/`,
shared helpers under `helpers/` (R14 LightGlue ownership), structured
JSON logging, runtime composition root with env-var fail-fast gate,
healthcheck module shared by Docker and CI smoke.
- CMake top-level + `cmake/{build_options,dependencies,strategies}.cmake`
with the BUILD_* per-binary flags (ADR-002) and pinned external git
refs for OKVIS2 / VINS-Mono / GTSAM / FAISS / OpenCV >=4.12.0.
- Three Dockerfiles (companion-tier1, operator-tooling,
mock-suite-sat-service) + two compose files (dev + Tier-1 test).
- Four GitHub Actions workflows: ci.yml (lint/unit/integration/dual
binary build/SBOM diff/security), ci-tier2.yml (self-hosted Jetson
AC-bound NFTs), release.yml, cve-rescan.yml.
- Two CI gate scripts: `ci/sbom_diff.py` (deployment SBOM subset +
R02 exclusion), `ci/opencv_pin_gate.py` (>=4.12.0 enforcement,
D-CROSS-CVE-1).
- Alembic-driven Postgres 16 initial migration `0001_initial.py`
mirroring satellite-provider tiles + flights + sector_classifications
+ manifests + engine_cache_entries (data_model.md s 2).
- Tier-1 test scaffolding: 95 passing unit tests covering every AC,
per-component smoke tests, structured logging JSON output check,
env-var gate check, healthcheck import check. Two CI-gated tests
(cmake configure, actionlint) skip locally with explicit reasons.
- Batch report + code review report under `_docs/03_implementation/`.
Verdict: PASS_WITH_WARNINGS (two Low findings, both informational).
Co-authored-by: Cursor <cursoragent@cursor.com >
2026-05-11 01:00:28 +03:00
Oleksandr Bezdieniezhnykh