Skip to content

Commit

Permalink
docs(how-to-guides): add training docs for centerpoint
Browse files Browse the repository at this point in the history
Signed-off-by: Kaan Çolak <[email protected]>
  • Loading branch information
kaancolak committed Nov 13, 2023
1 parent 7eb7942 commit c2f7b5b
Showing 1 changed file with 11 additions and 4 deletions.
Original file line number Diff line number Diff line change
Expand Up @@ -28,10 +28,17 @@ an example dataset containing three distinct classes (green, yellow, red), which

Detailed instructions for training the traffic light classifier model can be found **[here](https://github.com/autowarefoundation/autoware.universe/blob/main/perception/traffic_light_classifier/README.md)**.

<!--
## Training CenterPoint 3D object detection model

Training traffic light detection model and lidar CenterPoint model will be added there.
The CenterPoint 3D object detection model within the Autoware has been trained using the **[open-mmlab/mmdetection3d](https://github.com/open-mmlab/mmdetection3d)** repository. CenterPoint
implementation of mmdetection3d uses 10 input feature for PointPillars voxel encoder. However, Autoware employs 9 pillar features for CenterPoint, in accordance with the original research paper. Therefore, we have
fork the original repository and made the necessary code modifications to support the use of 9 PointPillar input features. The forked repository can be found **[here](https://github.com/autowarefoundation/mmdetection3d)**.

-->
To train custom CenterPoint models and convert them into ONNX format for deployment in Autoware, please refer to the instructions provided in the README file included with the
**"lidar_centerpoint"** package. These instructions will provide a step-by-step guide for training CenterPoint model.

## Training CenterPoint 3D object detection model
In order to assist you with your training process, we have also included an example dataset in the Tier4 dataset format.

Check warning on line 40 in docs/how-to-guides/training-machine-learning-models/training-models.md

View workflow job for this annotation

GitHub Actions / spell-check-differential

Unknown word (Tier4)
This dataset contains 600 lidar frames and covers 5 classes, including 6905 cars, 3951 pedestrians, 75 cyclists, 162 buses, and 326 trucks.
You can utilize this example dataset to facilitate your training efforts.

Detailed instructions for training the CenterPoint model can be found **[here](https://github.com/autowarefoundation/autoware.universe/blob/main/perception/lidar_centerpoint/README.md)**.

0 comments on commit c2f7b5b

Please sign in to comment.