crosstyan d86e9f11d7 fix(tracking): construct RPT camera payloads via typed bindings ...

Switch the triangulation adapter from ad hoc Python dictionaries to rpt.make_camera so the tracker passes the exact typed camera objects expected by the current RPT bindings.

The camera convention test now asserts against the bound camera object's numeric fields instead of raw dict payloads, which keeps the test aligned with the compiled interface.

2026-03-27 12:02:41 +08:00

..

support

feat(detection): add aligned video preparation helpers

2026-03-27 12:02:34 +08:00

__init__.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_actual_test_parquet.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_camera_conventions.py

fix(tracking): construct RPT camera payloads via typed bindings

2026-03-27 12:02:41 +08:00

test_detection_runner.py

fix(detection): preserve offline frames in the runner

2026-03-27 12:02:27 +08:00

test_detection_video_parquet.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_kinematics_seed.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_patch_mmdet.py

build: add reproducible mmdet patch script

2026-03-26 16:35:56 +08:00

test_protocol_and_normalization.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_tracker_single_person.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_tracker_smoke.py

feat!: reorganize detection and tracking pipeline

2026-03-26 16:24:27 +08:00

test_video_alignment.py

feat(detection): add aligned video preparation helpers

2026-03-27 12:02:34 +08:00