sergiomsilva / alpr-unconstrained

License Plate Detection and Recognition in Unconstrained Scenarios

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Does someone ever tried to convert this to TensorFlow Lite to make it work on Android ?

tomkisiela opened this issue · comments

I'm trying to convert the whole process to tensorflow lite but I'm a bit lost.
I first converted wpodnet and got this as input/output :
name: input type: float32[1,1,1,3] location: 0
name: Identity type: float32[1,1,1,8] location: 98
So if I understand well that means the input is a pixel right ?

Any success ?