Visualization tools are located in tools
directory. To visualize dataset and the corresponding prediction, try to
use viz_data.py
script.
Taking NuScenes dataset as an example, the data to be visualized should be formatted as a .txt
file, with each row being one of the objects:
obj. type | obj. code | score | x | y | z | l | w | h | angle | vx | vy |
---|---|---|---|---|---|---|---|---|---|---|---|
str | int | float | float | float | float | float | float | float | float | float | float |
- put all prediction files in one single folder (e.g., second_pred)
- run the command with point cloud file as argument, with
YOUR_PC_FILE_PATH
as the path to pc file andYOUR_PRED_PATH
is the path you specified in step 1:
python viz_data.py --cfg_file cfgs/viz_configs/nuscenes.yaml --data_path {YOUR_PC_FILE_PATH} --pred_path {YOUR_PRED_PATH}
This should give you the prediction with point cloud shown in a separate window as below: