Reading input as binary from Tensorflow C++ api

I have input images written in the binary files. I have to read binary file input and create a tiling of 512*512 image and do the inference and then padd all the output and return a single file.

I am using Tensorflow C++ API. I have the inference pipeline ready. I am struggling with reading binary files with TensorFlow and creating a tile and feed into the inference pipeline.

I have seen the available API. But nothing helps me to solve my problem. please guide How I can do it?


Can’t you read the binary file and do the tiling using regular C++ and just give the resulting image for the inference?