Open kkafieh opened 3 years ago
Try deleting all the cache files and running it again.
Out of Memory.
I deleted... -shinzo_abe_dsfd.pkl -shinzo_abe_dsfd_seq.pkl
Is there anything else I should have deleted? Not sure what the cache files are other than that.
It seemed to get much further. But now I get the Out of memory error below...
File "swap.py", line 504, in <module>
main(**vars(parser.parse_args()))
File "swap.py", line 498, in main
face_swapping(source[0], target[0], output, select_source, select_target)
File "swap.py", line 240, in __call__
target_cache_dir, target_seq_file_path, _ = self.cache(target_path)
File "C:\DATA\DEV\PROJECTS\fsgan\preprocess\preprocess_video.py", line 446, in cache
self.face_detector(input_path, det_file_path)
File "C:\DATA\DEV\PROJECTS\face_detection_dsfd\face_detector.py", line 92, in __call__
detections_batch = self.net(frame_tensor_batch)
File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "C:\DATA\DEV\PROJECTS\face_detection_dsfd\face_ssd_infer.py", line 415, in forward
sources[0] = self.cpm3_3(sources[0])
File "C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "C:\DATA\DEV\PROJECTS\face_detection_dsfd\face_ssd_infer.py", line 47, in forward
return torch.cat((x1_1, x2_1, x3_1), 1)
RuntimeError: CUDA out of memory. Tried to allocate 1.98 GiB (GPU 0; 12.00 GiB total capacity; 7.45 GiB already allocated; 0 bytes free; 9.52 GiB reserved in total by PyTorch)
I have the same issue. Did you solved it? @kkafieh Is it general issue? @YuvalNirkin
No. We gave up.
What kind of card so you have? We have a GeForce 3060.
I have Quadro RTX 5000
I am just trying to get this running for the first time. Running on RTX 3060
On trying my first swap with the example files I initially had an incompatibility message for my 3060, but installing pytorch v1.8.0 and cuda 11.3 fixed that. (conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge)
I then received an error: "RuntimeError: Cache file does not exist: ../docs/examples/shinzo_abe\shinzo_abe_dsfd_seq.pkl" Which I finally got passed by copying and renaming another pkl file. It seemed to allow the process to continue. although obviously its pretty sketchy.
Now I have this problem...
No "id" attribute? This looked like a programming error. I poked around in the code, but I am new to python and couldn't spot anything obvious. But also, the code should already be fine.
Is this an interpreter version problem? Somrthing else?
Python version 3.8.8 Pytorch version 1.8.0 Conda version 4.10.1 Cuda version 11.3.109