Visualize during test time

[Amol Agrawal]

Hi,

I was wondering if you could tell about how to visualize the result during test time. I know how to use ROS to visualize LIDAR data, but during test when the data starts being outputted to ROS, I am unable to see it in RViz.

[smanshei]

I think you can try to set reference frame

[smanshei]

rosrun tf static_transform_publisher 0 0 0 0 0 0 1 map velodyne 10

[Amol Agrawal]

hey @ansabsheikh9 thank you for replying. so based on your suggestion here is what i did.
used

    pcd_path = "../data/testing/velodyne/002397.bin"                                                                                                                                           
    calib_path = "../calib/testing/calib/002397.txt"                                                                                                                                           
    test(1, pcd_path, label_path=None, resolution=0.1, calib_path=calib_path, dataformat="bin", is_velo_cam=True, \                                                                            
        scale=8, voxel_shape=(800, 800, 40), x=(0, 80), y=(-40, 40), z=(-2.5, 1.5))       

to test the model.

started roscore , followed by python model_01_deconv.py followed by rosrun tf static_transform_publisher 0 0 0 0 0 0 1 map velodyne 10 followd by rviz

I then added a pointcloud to rviz and selected velodyne as fixed frame and /points_raw as topic in the pointckoud.

I am able to see a static pointcloud formation which changes if i change topic to /points_raw1

Am i doing something wrong? Also how do I add the identified cars to the visualization(through bounded boxes)

[smanshei]

@pfrcks I think every thing is alright,
You can change the color of points_raw1 and display both at the same time, so you will see this topic displays the detected bounding box corners i.e. 8 there are more than 8 points for one car so you have to implement NMS.

[Amol Agrawal]

hi @ansabsheikh9 thank you so much. it really cleared my doubts. :)

I have just one other query. There is no function definition for the the function filter_car_data, due to which I am unable to use train_test(). Any idea about this?

[smanshei]

#4 here you can have a look at the solution mentioned by @yukitsuji

[Amol Agrawal]

@ansabsheikh9 thank you so much for your replies. They have been very helpful to me.

I wanted to evaluate the model after training but I do not think that there is any code present in this repo which gives the accuracy of the model.
Do you have an idea as to how I could get the accuracy of the model?

[smanshei]

I think you have to write the code for this, you can use KITTI evaluation code, provided in dev kit aswell. If you find anything please update me aswell.