RuntimeError: storage has wrong size: expected 0 got 1728

[simdjeff]

Exception has occurred: RuntimeError
RealBasicVSR: PerceptualLoss: storage has wrong size: expected 0 got 1728

During handling of the above exception, another exception occurred:

File "D:\Users......\RealBasicVSR-master\realbasicvsr\models\restorers\real_basicvsr.py", line 65, in init
super().init(generator, discriminator, gan_loss, pixel_loss,

During handling of the above exception, another exception occurred:

File "D:\Users......\RealBasicVSR-master\realbasicvsr\models\builder.py", line 20, in build
return build_from_cfg(cfg, registry, default_args)
File "D:\Users......\RealBasicVSR-master\realbasicvsr\models\builder.py", line 58, in build_model
return build(cfg, MODELS, dict(train_cfg=train_cfg, test_cfg=test_cfg))
File "D:\Users......\RealBasicVSR-master\inference_realbasicvsr.py", line 67, in init_model
model = build_model(config.model, test_cfg=config.test_cfg)
File "D:\Users......\RealBasicVSR-master\inference_realbasicvsr.py", line 81, in main
model = init_model(args.config, args.checkpoint)
File "D:\Users......\RealBasicVSR-master\inference_realbasicvsr.py", line 149, in
main()

[simdjeff]

This error is caused by PerceptualLoss. If PerceptualLoss is removed, the error will disappear. However it's not easy to handle PerceptualLoss. The downloaded “vgg19-dcbb9e9d.pth” file is put into the .cache/torch/hub/checkpoints, but the build_model function didn't work either.

[ckkelvinchan]

Hi @simdjeff, since we currently support test only, PerceptualLoss should not be used. What command did you use?

The training code will be released next week.

[simdjeff]

Hi @ckkelvinchan. I used the command
"python inference_realbasicvsr.py configs/realbasicvsr_x4.py checkpoints/RealBasicVSR_x4.pth data/demo_000 results/demo_000".
Yes, PerceptualLoss can not be used for all test files with PerceptualLoss in mmediting/tests. I suppose the error doesn't occur only in ReadBasicVSR.
The error occurs in the function "build_from_cfg". The PerceptualLoss is used in eq.(8) in the paper "Investigating Tradeoffs in Real-World Video Super-Resolution". I wonder how to make RealBasicVSR work when building a model with PerceptualLoss?
Thank you. I am looking forward to the new training code.

[ckkelvinchan]

Are you using Windows?