I used my own data from a 16 layer Velodyne lidar for testing and it gave weird results?

[MohamedAboushnief]

I had some data collected at my university in a test track and the data was .pcd files containing the point cloud data and I converted each .pcd file into .bin file to fit in this ComplexYOLOV3 model and to be same format as KITTI velodyne data which every .bin file contains (x,y,z,reflectance) and I saved it into a directory to test on it in the ComplexYOLOV3 architecture which is in this repo and it gave me this weird results which is here in this picture.

can somebody tell me what to do?

[abhigoku10]

@MohamedAboushnief find the difference between the custom data and the kitti dataset , since the kitti dataset is capture from velodyne 64 sensor the points are more denser but in ur custom data they are a bit sparse based on that you can custom training

[MohamedAboushnief]

So if I train on my data that is 16 layers will the model work well? or should I change something?

[imdsafi09]

I also want to train my own dataset but the problem is I dont know which annotating tool i have to use to get the similar label as Kitti

[GUZ-lkc2018]

@MohamedAboushnief I have the same problem, have you solved it? Can you give me some advice

[MohamedAboushnief]

I didnt solve it :(

[BayaCherif]

you can use labelcloud or supervisely to create 3d bounding box data(length,width,height) + coordinates of the center of the object(x,y,z) and the y orientation.There are more info u need to write by yourself.Check the kitti documentation to understand more

I also want to train my own dataset but the problem is I dont know which annotating tool i have to use to get the similar label as Kitti