Closed camilo1704 closed 1 year ago
Hey! This is because the self.tail
of the model used for denoising and super-resolution is different. So, just set strict_load_g
to FLASE. Try the following command instead:
python infer.py -expe_opt options/repr/ESWT/ESWT-12-12_LSR.yml -task_opt options/task/Denoising.yml --force_yml path:strict_load_g=false
Note that this will replace self.tail
with randomly initialized, untrained layers. You can use the SIDD dataset or your own dataset to fine-tune ESWT to get better denoising results.
Hi, first at all great work guys!! I'm trying to use the denoising task and using the following command:
python infer.py -expe_opt options/repr/ESWT/ESWT-12-12_LSR.yml -task_opt options/task/Denoising.yml
got the next output:Do you have any advice? Thank you in advance.