Jesús Royeth 5a3e625e9d perf(inference): fuse common uint8 image conversion (#1390) ...

* perf(inference): fuse common uint8 image conversion
* test: cover one-channel uint8 predict dispatch


---------

Co-authored-by: Jesús Royeth <JESUSROYETH@users.noreply.github.com>
Co-authored-by: jirka <6035284+borda@users.noreply.github.com>
Co-authored-by: Codex <codex@openai.com>

2026-08-24 17:30:44 +02:00

..

__init__.py

refactor(tests): restructure tests/ to mirror src/rfdetr/ layout (#1166)

2026-06-26 21:37:01 +02:00

helpers.py

perf(inference): skip upsampling masks the caller's threshold discards (#1265)

2026-07-31 21:28:52 +02:00

test_from_checkpoint.py

Add doctests to helper functions in tests/evaluation and tests/inference (#1291)

2026-08-06 08:51:18 +02:00

test_inference_keypoints.py

refactor(tests): restructure tests/ to mirror src/rfdetr/ layout (#1166)

2026-06-26 21:37:01 +02:00

test_inference.py

refactor(tests): restructure tests/ to mirror src/rfdetr/ layout (#1166)

2026-06-26 21:37:01 +02:00

test_model_device_move.py

fix(inference): normalise the index-less cuda device before the deferred-move guard compares it (#1311)

2026-08-10 13:00:34 +02:00

test_model_inference.py

fix: guard ModelContext.reinitialize_detection_head against cleared weights (#1283)

2026-08-05 23:43:46 +02:00

test_predict_eval_mode.py

perf(inference): skip upsampling masks the caller's threshold discards (#1265)

2026-07-31 21:28:52 +02:00

test_predict.py

perf(inference): fuse common uint8 image conversion (#1390)

2026-08-24 17:30:44 +02:00

test_public_api_keypoints.py

fix(keypoints): normalize preview scores and ignore removed MLP weights (#1192)

2026-07-08 14:45:18 +02:00

test_supervision_editable_keypoints.py

refactor(tests): restructure tests/ to mirror src/rfdetr/ layout (#1166)

2026-06-26 21:37:01 +02:00

test_trt_inference.py

fix(export): pass CUDA device to ONNX Runtime benchmark (#1346)

2026-08-15 22:32:18 +02:00