Open z-hou opened 5 years ago
I haven't met this. One reason might be the out-of-memory issue. The default ddd model takes the original input resolution. You can try specifying a lower resolution by --input_h xxx --input_w xxx
.
I can get multi_pose result by using demo.py with some modification. command: python demo.py mul_tipose --demo imagefolder --loadmodel ../models/multi_pose_hg_3x.pyh --arch hourglass but when I make the same change to get the 3d-detection result, command: python demo.py ddd --demo imagefolder --loadmodel ../models/ddd_3dop.pth without specify the --arch augment, I got this error: THcudaCheck FAIL file=/pytorch/sten/src/THC/THCgeneral.cpp line=383, error=11: invalid argument and Runtime error: cublas runtime error: the GPU program failed to execute at /pytorch/aten/src/THC/THCBlas.cu:425 I wonder how can fix this problem, thanks very much
I have met same problem, have you solved it?
I can get multi_pose result by using demo.py with some modification. command: python demo.py mul_tipose --demo imagefolder --loadmodel ../models/multi_pose_hg_3x.pyh --arch hourglass but when I make the same change to get the 3d-detection result, command: python demo.py ddd --demo imagefolder --loadmodel ../models/ddd_3dop.pth without specify the --arch augment, I got this error: THcudaCheck FAIL file=/pytorch/sten/src/THC/THCgeneral.cpp line=383, error=11: invalid argument and Runtime error: cublas runtime error: the GPU program failed to execute at /pytorch/aten/src/THC/THCBlas.cu:425 I wonder how can fix this problem, thanks very much