You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/docs/04-typescript-api/02-computer-vision/PoseEstimationModule.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -96,7 +96,7 @@ detections[0].THUMB_TIP; // { x, y }
96
96
97
97
The `.pte` binary must expose a `forward` method (or per-input-size methods such as `forward_384`, `forward_512`, `forward_640` for multi-resolution models) with the following interface:
98
98
99
-
**Input:** one `float32` tensor of shape `[1, 3, H, W]` — a single RGB image, values in `[0, 1]` after optional per-channel normalization `(pixel − mean) / std`. H and W are read from the model's declared input shape at load time. The mean and std vectors are supplied via `preprocessorConfig.normMean` and `preprocessorConfig.normStd` on the [`PoseEstimationConfig`](../../06-api-reference/interfaces/PoseEstimationConfig.md) you pass to `fromCustomModel`; if omitted, the runtime feeds the resized image without normalization.
99
+
**Input:** one `float32` tensor of shape `[1, 3, H, W]` — a single RGB image, values in `[0, 1]` after optional per-channel normalization `(pixel − mean) / std`. H and W are read from the model's declared input shape at load time. The mean and std vectors are supplied via `preprocessorConfig.normMean` and `preprocessorConfig.normStd` on the [`PoseEstimationConfig`](../../06-api-reference/type-aliases/PoseEstimationConfig.md) you pass to `fromCustomModel`; if omitted, the runtime feeds the resized image without normalization.
100
100
101
101
**Outputs:** exactly three `float32` tensors, in this order:
Copy file name to clipboardExpand all lines: packages/react-native-executorch/common/rnexecutorch/tests/README.md
+4-1Lines changed: 4 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -69,5 +69,8 @@ To add new test you need to:
69
69
LIBS opencv_deps
70
70
)
71
71
```
72
-
* Lastly, add the test executable name to the run_tests script along with all the needed URL and assets.
72
+
* In `run_tests.sh`:
73
+
* Add the test executable name to `TEST_EXECUTABLES`.
74
+
* Add any models/files the test downloads at runtime to `MODELS` (filename + URL), **and** register every downloaded file the test loads in the `models_for_test()` case statement. The runner pushes only the files listed there from `$MODELS_DIR` to the device for that test, runs it, and removes them afterwards — anything missing won't be on the device when the test runs. Tests with no model dependencies don't need an entry.
75
+
* Repo-bundled fixtures (small images, audio, etc.) go in `TEST_ASSETS` instead. Those are pushed once up front and stay on the device; do not list them in `models_for_test()`.
0 commit comments