When I walk a directory recursively, it is tacking on an additional non-existant file from one of the subdirectories.  Here's the code:

    def copy_dir(self, src, dest):
        result = sftp.mkdir(dest)
        for dirname, dirnames, filenames in os.walk(src):
            for subdirname in dirnames:
                print("entering dir:" + subdirname)
                self.copy_dir(os.path.join(src, subdirname), os.path.join(dest, subdirname))
            for filename in filenames:
                print("copying:" + filename)

Here's the output:

entering dir:include
copying:CallbackInterface.h
copying:ClockInterface.h
entering dir:tcc
copying:CallbackInterface.tcc
copying:CMakeLists.txt
copying:darknet.i
copying:darknet.i.h
copying:darknet.obj
copying:darknet.py
copying:darknetImageNetLabels.txt
copying:darknetPYTHON_wrap.cxx
copying:darknetPYTHON_wrap.h
copying:darknet_config.json
copying:demo.py
copying:demoHelper.py
copying:OpenBLASSetup.cmake
copying:runtest.sh
copying:schoolbus.png
copying:CallbackInterface.h
copying:ClockInterface.h
copying:CallbackInterface.tcc

The last 3 files listed here doesn't exist, they are a repeat of the files found in the subdirectories.

Read the documentation of os.walk() again. It already walks the complete directory tree starting with src. 

When you invoke it again by calling your copy_dir() method recursively you will of course see once more the files and directories in the respective subdirectory.

Oh, my bad then. Apologies for the noise in your system.