Download Link in Baidu Cloud Disk password: aj7z
👆 includes the re-annotated labels and prompts.
Please download the VoD dataset and organize the dataset like this:
- VoD
|
--- lidar
|
--- training
|
--- label_2
--- prompt
--- image_2
--- radar_5frames
|
--- training
|
--- label_2
--- prompt
--- image_2
If you are familiar with mmdetection3D framework, you can work on it fast.
You should use the official evaluation code of VoD to evaluate the predicted results of Talk2Radar.