A summary and list of open source 3D multi object tracking and datasets at this stage.
Solution | Popularity | Feature | Paper | Language | Sensors | Dataset |
---|---|---|---|---|---|---|
AB3DMOT (IROS 2020) | The baseline of a lot of 3D MOT framework since 2019 | here | python | LIDAR | KITTI | |
mahalanobis_3d_mot (2019) | use mahalanobis distance based on AB3DMOT (the first place award in NuScenes Tracking Challenge) | here | python | LIDAR | KITTI | |
3D-Multi-Object-Tracker | Fast (700FPS), support online, near online and global implementation in 3D MOT, including visualization code | python | LIDAR | KITTI Waymo | ||
EagerMOT (ICRA 2021) | Improve the online 3D multi-object tracking performance by using 2D detections to support tracking when 3D association fails. | here | python | LIDAR + camera | KITTI Nuscenes | |
CenterPoint (2021) | 3D Object Detection and Tracking using center points in the bird-eye view | here | python | Waymo Nuscenes | ||
Monocular Quasi-Dense 3D Object Tracking (QD-3DT ,2021) website |
an online framework detects and tracks objects in 3D using quasi-dense object proposals from 2D images. | here | python | camera | NuScenes KITTI Waymo | |
Detection and Tracking of Moving Objects with 2D LIDAR(2017) | Tracking moving objects using L-Shape feature from 2D LiDAR | here | c++(ros) | 2d LIDAR | own |
Dataset | Tutorial | Paper |
---|---|---|
KITTI | ||
NuScenes | tool | |
Waymo | ||
Apollo Scape | github | here |
H3D Dataset | homepage | here |
For specific parameters and codes for MOT accuracy evaluation, I recommend the following two githubs:
If you are also interested in this topic, please feel free to discuss! 😄 [Contact Me](zhen tan: [email protected])
希望找到志同道合的小伙伴~