FairMOT icon indicating copy to clipboard operation
FairMOT copied to clipboard

UserWarning: Error checking compiler version for cl:(dcnv2)

Open DioMou opened this issue 3 years ago • 2 comments

@ifzhang I am having problems with setup the dcnv2, it gives me the following warning if I run python setup.py build develop. Although the building process successfully finished, but if I try to import _ext, it says DLL load failed while importing _ext 动态链接库(DLL)初始化例程失败(dll initlization failed), so I suspect this is because of the user warning below. running build running build_ext cl(it says this is my printed compiler name but I am not running it on visual studio, this is simply vscode with python 3.8 torch 3.7.0) Traceback (most recent call last):(self printed traceback, start here) File "C:\Users\LXAB\anaconda3\envs\mltorch\lib\site-packages\torch\utils\cpp_extension.py", line 270, in check_compiler_abi_compatibility compiler_info = subprocess.check_output(compiler, stderr=subprocess.STDOUT) File "C:\Users\LXAB\anaconda3\envs\mltorch\lib\subprocess.py", line 415, in check_output return run(*popenargs, stdout=PIPE, timeout=timeout, check=True, File "C:\Users\LXAB\anaconda3\envs\mltorch\lib\subprocess.py", line 493, in run with Popen(*popenargs, **kwargs) as process: File "C:\Users\LXAB\anaconda3\envs\mltorch\lib\subprocess.py", line 858, in __init__ self._execute_child(args, executable, preexec_fn, close_fds, File "C:\Users\LXAB\anaconda3\envs\mltorch\lib\subprocess.py", line 1311, in _execute_child hp, ht, pid, tid = _winapi.CreateProcess(executable, args, FileNotFoundError: [WinError 2] 系统找不到指定的文件。(means system cannot find file,ends here) C:\Users\LXAB\anaconda3\envs\mltorch\lib\site-packages\torch\utils\cpp_extension.py:278: UserWarning: Error checking compiler version for cl: [WinError 2] 系统找不到指定的文件。 warnings.warn('Error checking compiler version for {}: {}'.format(compiler, error)) building '_ext' extension I did try to change decode to gbk but does not work. Anybody can help? I am using python 3.8, torch vision 0.8, cudatoolkit 10.2,pytorch 1.7

DioMou avatar Jul 28 '21 05:07 DioMou

请问你解决这个问题了吗,我也遇到了同样的问题

DEEPMINDing avatar Sep 30 '21 10:09 DEEPMINDing

请问你解决这个问题了吗,我也遇到了同样的问题

没有,我认为有可能是系统的问题,dcn可能得在linux上才能弄,windows我认为可能有一些谦容问题。The original dcnv2 module specified that it is developed under linux environment. WS2可能能够解决,但需要重新安装cuda等一系列操作,我没有完全尝试。要重新train的话我试了下应该只有hrnet可以直接用

DioMou avatar Sep 30 '21 21:09 DioMou