tensorflow-qtcreator-cppApi

Run tensorflow using cpp_api on qtcreator to predict the score of image using tensorflow prebuilt binary on windows 10(64-bit)

Description:

The program is used to load a tensorflow model (.pb) to predict the score of images in a folder using tensorflow_cpp api.

Pre-requisites:

1. Tensorflow Binary(Here 1.8 binary have been used): Pre-built tensorflow binary can be downloaded from the following location: https://github.com/fo40225/tensorflow-windows-wheel/tree/master/1.8.0/cpp.
   Note:
   Here i have used binary build using sse2 as i was finding linking issues in avx2 for my system which i have to still verify.

2. Tensorflow Model (.pb): Tensorflow model trained in python have used. The model was initially build using Keras and then converted into .pb for using it in tensorflow.
   Note:
   Note the input and output node of your model for giving input to the program

3. Opencv-Release files: Link the opencv release lib files to the program. The tensorflow lib is also built in release mode.

4. MSVC compiler: Inorder to run the program we have to use MSVC compiler which have also been defined in the program.

5. Please refer the pro file for more details.

Steps to Run the program:

Open the program in qt-creator. Give the path of image folder and tensorflow model. Build the program. As this program is build in release mode include the necessary dll files in the release folder.
Note: Please copy paste all the necessary dll files required in the release folder during running the program