MV-FCOS3D++
| Authors | Tai Wang*, Qing Lian*, Chenming Zhu, Xinge Zhu, Wenwei Zhang |
| Description | We build a multi-view framework with temporal stereo modeling to convert multi-view features to a 3D grid space and perform 3D detection thereon. The ResNet101-DCN backbone based on FCOS3D++ is pretrained on Waymo with only object annotations. We do not involve lidar depth labels both during training and inference. Code will be released at MMDetection3D. |
| Project Link | Link |
TYPE_VEHICLE
- Sensors: C
TYPE_PEDESTRIAN
- Sensors: C
TYPE_CYCLIST
- Sensors: C
ALL_NS
- Sensors: C