Tensorflow C++ preparing vector input and running inference on a model

I am trying to build an inference application using TensorFlow in C++ to be used in an optimization program. I have already built and trained the CNN model using python, and have been able to load it into C++ as a pb file, but I haven’t been able to figure out how to run an inference with my Input type, which is a multidimensional floating number vectors. That’s where you come in! What needs to be done can be described in the following steps -Pass the loaded model and tensorflow session into an inference function -convert input vectors into tensors -perform an inference on the loaded model with the input tensors -convert the output tensors into a vector Attached is the code for the project and one of the models you will have to run it with. Properly completing the cnn_inference_regression, and cnn_inference categorization functions is the scope of the project. See the string ‘#wrk’ in the cpp file for all points of interest. The project is similar to the tutorial below except you are handling the reading and output data as I have specified. [login to view URL] This shouldn’t be a hard task if you have experience and knowledge with TensorFlow's c++ implementation, which I lack. Ideally, I would like to work with someone who can recreate similarly to my platform which is described by the following. This is not required, but final code is expected to work on this platform. -Visual Studio 2017 15.6 -Tensorflow v1.4 -Windows 10 -CUDA 8 -cuDNN 6 I have built my Tensorflow library according to this link following the path with the gpu implimentation. [login to view URL]