zllrunning/video-object-removal

undefined symbol: __cudaRegisterFatBinaryEnd if run demo.py

yaroslavmavliutov opened this issue · 3 comments

Hello! When I try ro run demo.py , i obtain an error

Traceback (most recent call last):
File "demo.py", line 3, in
from inpaint import inpaint
File "/content/video-object-removal/inpaint.py", line 5, in
from inpainting.model import generate_model
File "/content/video-object-removal/inpainting/model.py", line 3, in
from inpainting.models import vinet
File "/content/video-object-removal/inpainting/models/vinet.py", line 6, in
from inpainting.models.flow_modules import (WarpingLayer, LongFlowNetCorr, MaskEstimator_ )
File "/content/video-object-removal/inpainting/models/flow_modules.py", line 7, in
from inpainting.models.correlation_package.modules.correlation import Correlation
File "/content/video-object-removal/inpainting/models/correlation_package/modules/correlation.py", line 3, in
from ..functions.correlation import CorrelationFunction
File "/content/video-object-removal/inpainting/models/correlation_package/functions/correlation.py", line 3, in
from .._ext import correlation
File "/content/video-object-removal/inpainting/models/correlation_package/_ext/correlation/init.py", line 3, in
from ._correlation import lib as _lib, ffi as _ffi
ImportError: /content/video-object-removal/inpainting/models/correlation_package/_ext/correlation/_correlation.so: undefined symbol: __cudaRegisterFatBinaryEnd

I used pytorch 0.4.0, cuda 8.0 (tried 9.0 also)

I met this problem under pytorch 0.4.0, python 3.5.6, cuda 11.1, gcc 7.6 and solved it by changing environment to pytorch 0.4.0, python 3.5.6, cuda 9.0, gcc 5.5.0. After changing env, rerun .sh files to compile dependencies. Hope it useful.

I met this problem under pytorch 0.4.0, python 3.5.6, cuda 11.1, gcc 7.6 and solved it by changing environment to pytorch 0.4.0, python 3.5.6, cuda 9.0, gcc 5.5.0. After changing env, rerun .sh files to compile dependencies. Hope it useful.
that is useful, gcc 5.4 also works!