From f267ef2790be2f69bd013982db922e7a5bc41d16 Mon Sep 17 00:00:00 2001 From: kminoda <44218668+kminoda@users.noreply.github.com> Date: Wed, 31 Jul 2024 12:17:24 +0900 Subject: [PATCH] docs(centerpoint): add description for ml package params (#8187) --- perception/autoware_lidar_centerpoint/README.md | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/perception/autoware_lidar_centerpoint/README.md b/perception/autoware_lidar_centerpoint/README.md index 83c18b76fd079..471712a637dec 100644 --- a/perception/autoware_lidar_centerpoint/README.md +++ b/perception/autoware_lidar_centerpoint/README.md @@ -28,6 +28,22 @@ We trained the models using . ## Parameters +### ML Model Parameters + +Note that these parameters are associated with ONNX file, predefined during the training phase. Be careful to change ONNX file as well when changing this parameter. Also, whenever you update the ONNX file, do NOT forget to check these values. + +| Name | Type | Default Value | Description | +| -------------------------------------- | ------------ | ------------------------------------------------ | --------------------------------------------------------------------- | +| `model_params.class_names` | list[string] | ["CAR", "TRUCK", "BUS", "BICYCLE", "PEDESTRIAN"] | list of class names for model outputs | +| `model_params.point_feature_size` | int | `4` | number of features per point in the point cloud | +| `model_params.max_voxel_size` | int | `40000` | maximum number of voxels | +| `model_params.point_cloud_range` | list[double] | [-76.8, -76.8, -4.0, 76.8, 76.8, 6.0] | detection range [min_x, min_y, min_z, max_x, max_y, max_z] [m] | +| `model_params.voxel_size` | list[double] | [0.32, 0.32, 10.0] | size of each voxel [x, y, z] [m] | +| `model_params.downsample_factor` | int | `1` | downsample factor for coordinates | +| `model_params.encoder_in_feature_size` | int | `9` | number of input features to the encoder | +| `model_params.has_variance` | bool | `false` | true if the model outputs pose variance as well as pose for each bbox | +| `model_params.has_twist` | bool | `false` | true if the model outputs velocity as well as pose for each bbox | + ### Core Parameters | Name | Type | Default Value | Description |