Visualize 3D Bounding Boxes Over Images and Lidar Data in ROS rviz

[tianweiy]

Visualize 3D Bounding Boxes Over Images and Lidar Data in ROS rviz

[tianweiy]

inference some of the pre-trained models provided in the MODEL ZOO on the test dataset of nuScenes and generate the co-ordinates of the bounding box for cars, pedestrians etc.

For the test, you first need to generate the info file using create_data.py and the version should be v1.0-test. After that, you do the inference in the standard way with dist_test.py by running

python tools/dist_test.py CONFIG_NAME --work_dir SOME_PATH  --checkpoint SOME_PATH --speed_test --testset 

And then how can we visualize them by overlaying them over camera images (front, front right, front left, back, back right and back left) and also in the point cloud visualization using ros rviz

I never use ros rviz in the past. To draw on the camera images using matplotlib. you can modify the nuscenes-devkit's help function Link.

I guess This script will also be helpful if you want to use some ros libarary.

[tianweiy]

You can use. But it will only visualize objects in Bird eye view. The devkit function also shows camera view. Yes, just change the path in demo.py and also in other configuration files change the data path accordingly. It is the reference frame coordinate (coordinate of the top lidar)

[iamsiddhantsahu]

@tianweiy Thanks! The same way demo.py uses the visual() function defined in demo_utils.py, can we not use this function to visualize?
Additionally, I wanted to inquire can the demo.py file be used with other configurations and pre-trained models? And the detected points in the demo.py file are with respect to which co-ordinate system?

[iamsiddhantsahu]

This is addressed in much details in #19