Waymo_Kitti_Adapter

This is a tool converting Waymo open dataset format to Kitti dataset format. This tool was modified from the following repositories:

Original Repository: https://github.com/Yao-Shao/Waymo_Kitti_Adapter Forked Repository: https://github.com/RocketFlash/Waymo_Kitti_Adapter

Instruction

1. Install pip packages

pip install -r requirements.txt

2. Download the Waymo dataset. Create a Waymo folder on your machine and organize the downloaded files as follows:

├── Waymo
│   ├── original
│   │   │──training
│   │   │   ├──training_0000
│   │   │   │   ├─.tfrecord files
│   │   │   ├──training_0001
│   │   │   │   ├─.tfrecord files
│   │   │   ├──...
│   │   │──validation
│   │   │   ├──validation_0000
│   │   │   │   ├─.tfrecord files
│   │   │   ├──validation_0001
│   │   │   │   ├─.tfrecord files
│   │   │   ├──...
│   │   │──testing
│   │   │   ├──testing_0000
│   │   │   │   ├─.tfrecord files
│   │   │   ├──testing_0001
│   │   │   │   ├─.tfrecord files
│   │   │   ├──...

3. Clone the waymo open dataset repo and follow the instructions on its Quick Start Page in order to build and test it.
4. Clone this repo to your computer, then copy the files from this repo to the waymo-od repo

cp Waymo-Kitti-Adapter/protocol_buffer/* waymo-od/waymo_open_dataset
cp Waymo-Kitti-Adapter/adapter.py waymo-od/
cp Waymo-Kitti-Adapter/adapter_lib.py waymo-od/

5. Copy adapter.py to waymo-od folder and set up the (our recommendation) following folder structure:

├── Waymo
│   ├── original
│   │   │──training
│   │   │   ├──training_0000, ...
│   │   │──validation
│   │   │   ├──validation_0000, ...
│   │   │──testing
│   │   │   ├──testing_0000...
│   ├── adapted
│   │   │──training
│   │   │──validation
│   │   │──testing

Create the adapted folder in Waymo directory with training, validation and testing folders inside. These will be locations that the adapter will save to. 6. Run adapter.py to save training data. Before running, open adapter.py and change the configurations at the top so that it suits to your own computer's path. Set DATA_PATH to be '{YOUR PATH}/Waymo/original/training' and KITTI_PATH to be '{YOUR PATH}/Waymo/adapted/training'.

python adapter.py

7. Run adapter.py to save validation data. Change DATA_PATH to be '{YOUR PATH}/Waymo/original/validation' and KITTI_PATH to be '{YOUR PATH}/Waymo/adapted/validation'.
8. Run adapter.py to save testing data. Change DATA_PATH to be '{YOUR PATH}/Waymo/original/testing' and KITTI_PATH to be '{YOUR PATH}/Waymo/adapted/testing'. Add the config option --test True.

python adapter.py --test True

9. Once completed, the folder tree will look like this:

...
├── Waymo
│   ├── original
│   │   │──training & testing & validation
│   ├── adapted
│   │   │──training
│   │   │   ├──calib & velodyne & label_0 & image_0
│   │   │──testing
│   │   │   ├──calib & velodyne & label_0
│   │   │──testing
│   │   │   ├──calib & velodyne & label_0 & image_0

Data specification

Cameras

Waymo dataset contains five cameras:

FRONT = 0;
FRONT_LEFT = 1;
FRONT_RIGHT = 2;
SIDE_LEFT = 3;
SIDE_RIGHT = 4;

all the names below with post-fix 0-4 is corresponding to these five cameras.

Label

label_0 to label_4 contains label data for each camera and label_all fonder contain all the labels.

All in vehicle frame.

For each frame, here is the data specification:

#Values    Name      Description
----------------------------------------------------------------------------
   1    type         Describes the type of object: 'Car', 'Van', 'Truck',
                     'Pedestrian', 'Person_sitting', 'Cyclist', 'Tram',
                     'Misc' or 'DontCare'
   1    truncated    Float from 0 (non-truncated) to 1 (truncated), where
                     truncated refers to the object leaving image boundaries
   1    occluded     Integer (0,1,2,3) indicating occlusion state:
                     0 = fully visible, 1 = partly occluded
                     2 = largely occluded, 3 = unknown
   1    alpha        Observation angle of object, ranging [-pi..pi]
   4    bbox         2D bounding box of object in the image (0-based index):
                     contains left, top, right, bottom pixel coordinates
   3    dimensions   3D object dimensions: height, width, length (in meters)
   3    location     3D object location x,y,z in camera coordinates (in meters)
   1    rotation_y   Rotation ry around Y-axis in camera coordinates [-pi..pi]
   1    camera_num	the camera number which the object belongs to, only exist
                     in label_all

Calib

P0-P4 : intrinsic matrix for each camera
R0_rect : rectify matrix
Tr_velo_to_cam_0 - Tr_velo_to_cam_4 : transformation matrix from vehicle frame to camera frame

Image

image_0 - image_4 : images for each

Lidar

Point cloud in vehicle frame.

x y z intensity

For more details, see readme.txt by KITTI.

References

1. Waymo open dataset
2. argoverse kitti adapter