[Question]The error of intall nndetection

[jk-blacker]

:question: Question

Hi! Awesome work :)

Recently, I tried to install nnDetection with pip install -e . , but I got a problem:

ERROR: Command errored out with exit status 1: command: /home/sensetime/anaconda3/envs/nndet/bin/python -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/home/sensetime/Documents/FHwork/code/nnDetection-main/setup.py'"'"'; file='"'"'/home/sensetime/Documents/FHwork/code/nnDetection-main/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(file) if os.path.exists(file) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' develop --no-deps cwd: /home/sensetime/Documents/FHwork/code/nnDetection-main/ Complete output (106 lines): ...... ERROR: Command errored out with exit status 1: /home/sensetime/anaconda3/envs/nndet/bin/python -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/home/sensetime/Documents/FHwork/code/nnDetection-main/setup.py'"'"'; file='"'"'/home/sensetime/Documents/FHwork/code/nnDetection-main/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(file) if os.path.exists(file) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, file, '"'"'exec'"'"'))' develop --no-deps Check the logs for full command output.

Thank you for your help!

[mibaumgartner]

Hi @jk-blacker ,

please run python utils.py (with the activated environment) from the scripts folder of nnDetection to collect additional information of your environment. Since I didn't encounter the mentioned error, something inside the environment is probably the reason.

Best, Michael

[YuqingCai0316]

I have the same problem,I have run python utils.py，and get Are there any mistakes?

[mibaumgartner]

Hi @YuqingCai0316 ,

thanks for the feedback. The configuration looks good in general. I'll try to reproduce the issue as soon as I can, due to an upcoming conference it will probably take some time though (sorry).

In the meantime, I would recommend using the docker installation if possible.

Best, Michael

[JoeNajm]

Hello @mibaumgartner @jk-blacker , I have the same issue, did you manage to find a solution for that problem? Thanks in advance ! Joe

[mibaumgartner]

Dear @JoeNajm ,

this error never occurred to me, so it is quite hard to diagnose it right now.

I think my candidates would be: [All of them are somewhat unlikely but assuming your environment is correct I'm not sure what else could cause this]

• having a cuda capable device present during installation, there were errors on HPC clusters
• if CUDA_HOME is not None, it might point to the wrong version even though the nvcc commands reports a suitable CUDA version
• there might be a mismatch between the CUDA and GCC version (https://gist.github.com/ax3l/9489132)
• maybe something related to the SM86 architecture which I didn't test yet

Best, Michael