Make it work with cpu only

[apple2373]

Nice work! It's very light (only 4MB of weights) tool to detect hands! I don't have cuda on my laptop and so wanted to run without cuda. I made it, but I had to modify nms to compile without gpu and also interference.

I basically did two modifications. Remove gpu compile code from utils/build.py

import os
from os.path import join as pjoin
import numpy as np
from distutils.core import setup
from distutils.extension import Extension
from Cython.Distutils import build_ext


def find_in_path(name, path):
    "Find a file in a search path"
    # adapted fom http://code.activestate.com/recipes/52224-find-a-file-given-a-search-path/
    for dir in path.split(os.pathsep):
        binpath = pjoin(dir, name)
        if os.path.exists(binpath):
            return os.path.abspath(binpath)
    return None


# Obtain the numpy include directory.  This logic works across numpy versions.
try:
    numpy_include = np.get_include()
except AttributeError:
    numpy_include = np.get_numpy_include()


def customize_compiler_for_nvcc(self):
    """inject deep into distutils to customize how the dispatch
    to gcc/nvcc works.

    If you subclass UnixCCompiler, it's not trivial to get your subclass
    injected in, and still have the right customizations (i.e.
    distutils.sysconfig.customize_compiler) run on it. So instead of going
    the OO route, I have this. Note, it's kindof like a wierd functional
    subclassing going on."""

    # tell the compiler it can processes .cu
    self.src_extensions.append('.cu')

    # save references to the default compiler_so and _comple methods
    default_compiler_so = self.compiler_so
    super = self._compile

    # now redefine the _compile method. This gets executed for each
    # object but distutils doesn't have the ability to change compilers
    # based on source extension: we add it.
    def _compile(obj, src, ext, cc_args, extra_postargs, pp_opts):
        print(extra_postargs)
        if os.path.splitext(src)[1] == '.cu':
            # use the cuda for .cu files
            self.set_executable('compiler_so', CUDA['nvcc'])
            # use only a subset of the extra_postargs, which are 1-1 translated
            # from the extra_compile_args in the Extension class
            postargs = extra_postargs['nvcc']
        else:
            postargs = extra_postargs['gcc']

        super(obj, src, ext, cc_args, postargs, pp_opts)
        # reset the default compiler_so, which we might have changed for cuda
        self.compiler_so = default_compiler_so

    # inject our redefined _compile method into the class
    self._compile = _compile


# run the customize_compiler
class custom_build_ext(build_ext):
    def build_extensions(self):
        customize_compiler_for_nvcc(self.compiler)
        build_ext.build_extensions(self)


ext_modules = [
    Extension(
        "nms.cpu_nms",
        ["nms/cpu_nms.pyx"],
        extra_compile_args={'gcc': ["-Wno-cpp", "-Wno-unused-function"]},
        include_dirs=[numpy_include]
    ),
]

setup(
    name='mot_utils',
    ext_modules=ext_modules,
    # inject our custom trigger
    cmdclass={'build_ext': custom_build_ext},
)

and comment out this line.

hand-detection.PyTorch/utils/nms_wrapper.py

Line 9 in ed1398d

from .nms.gpu_nms import gpu_nms

Then the test.py will work only with cpu.

[harsh-wardhan]

I'm getting this error on compilation:

nms/cpu_nms.c:8431:13: error: too many arguments to function ‘(PyObject * ()(PyObject , PyObject * const, Py_ssize_t))meth’
return (((__Pyx_PyCFunctionFast)meth)) (self, args, nargs, NULL);