Explanation for detect_lp() two input params which are hardcoded

[cassini-fly]

Hi,

Great work on this repo and thanks a lot for the content! I can see that in license-plate-detection.py:46 where detect_lp() is being called, there are two hard coded input params to this function.
Llp,LlpImgs,_ = detect_lp(wpod_net,im2single(Ivehicle),bound_dim,2**4,(240,80),lp_threshold)
Could we find any documentations on them? I see they correspond to 'net_step' and 'out_size'. I would like to know how these should be changed for different input image sizes?
Will keeping them as is, effect the execution time given different image sizes?