Training on custom long range dataset

[RocketFlash]

Hi @HuangJunJie2017 ! Thank you for amazing work!

I have a question about how to configurate model training parameters on my own data. I already prepared data and can start training, but I don't understand how properly configure grid_config and other parameters to make model converge. I have data from 3 cameras with different focal length

Objects are located in range 0-1000 meters from lidar. How to properly set grid_config ?

grid_config = {
'x': [0, 1024, 8],
'y': [-51.2, 51.2, 0.8],
'z': [-5, 3, 8],
'depth': [1.0, 60.0, 1.0],
}

is this config okay ?

[RocketFlash]

How to set 'depth' properly ?