Toolkit for JRDB dataset
This repo contain code for JRDB dataset and benchmark. For tracking evaluation, our code is adapted from TrackEval. With addition code from 3D tracking and OSPA metric. For detection evaluation, the code is adapted from Kitti Detection. For action evaluation, the code is adapted from AVA dataset. Please preceed to sub directory for specific instructions.
The visualisation is based on KITTI visualisation from Kuixu, we did some modifications (still developing).
Please note these coordinate system changes before generating submissions
Coordinates in LiDAR:
Coordinates in camera:
up z z front
^ x front /
| / /
| / 0 ------> x right
left y <------ 0 |
|
v
down y
The relative coordinate of bottom center in a LiDAR box is (0.5, 0.5, 0), and the yaw is around the z axis, thus the rotation axis=2.
Groundtruth box's coordinates:
Box in LIDAR
Box in camera:
/-------------/| /-------------/|
/| / | /| / |
/_____________/ | /_____________/ |
| | | | | | | |
| | z| | | | | | |
| | | /x | | | | | |
| | |/ | | | | | |
| |y<---+ | | | | | |
| | | | | | /z | |
| /+----------+--/ | /+-----/----+--/
|/ | / |/ +--->x| /
|_____________|/ |_______|_____|/
|y
x, y, z, l (x_size), w (y_size), h (z_size) x, y, z, l (z_size), w (x_size), h (y_size)
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent