Week 3

I have been integrating keras in smart-traffic-sensor. For this, I have to embeded Python code in C++. You can find information in the following links: 1 , 2 , 3. You can see a basic example in the next code (hello_python.cpp):

#include <stdio.h>
#include <python3.5/Python.h>

int main()
{
	PyObject* pInt;
	Py_Initialize();
	PyRun_SimpleString("print('Hello World from Embedded Python!!!')");
	Py_Finalize();
	printf("\nPress any key to exit...\n");
}

You have to compile in the following way:

gcc hello_python.cpp -o hello_python -L/usr/lib/python2.7/config/ -lpython2.7

The result is:

Hello World from Embedded Python!!!
Press any key to exit...

At the moment, I have used a COCO model(VGG_300_300_coco.h5 -> http://jderobot.org/store/deeplearning-networks/Keras/) for keras in a traffic video. At the moment I have used a coconut model for keras in a traffic video. Although I still have some conflict with the part of tensorflow already integrated, because in the part of keras some tensorflow function is also used and I am getting some problems by leaving both parts enabled. In the next video, you can see an example: