search

hellloxiaotian / ESRGCNN

Image Super-resolution with An Enhanced Group Convolutional Neural Network (Neural Networks, 2022)
https://github.com/hellloxiaotian/ESRGCNN
61 stars 5 forks source link

about blind image SR #8

Open CodeMadUser opened 2 years ago

CodeMadUser commented 2 years ago

hello PhD. TIAN,it's an amazing work.But when I tested the blind image super-resolution, I didn't get the same results as the low-resolution image.My folder directory is as follows: image

I used this directive：python tcw_sample_b.py --model esrgcnn --test_data_dir dataset/Set5 --scale 2 --ckpt_path checkpoint/esrgcnn.pth --sample_dir samples_singlemodel_urban100_x2 the result is: image I didn't get the results for img_006_SRF_2_LR.png.I don't know why,I'm looking forward to your replying.thank you. How do I test blind graphics for super resolution?

hellloxiaotian commented 2 years ago

Thank you for your concern. There is only a single image called img_006_SRF_2_LR.png, but the corresponding image img_006_SRF_2_HR.png is missing. So the dataset can't load the img_006_SRF_2_LR.png.

CodeMadUser commented 2 years ago

Thank you for your reply. And that is that this code is temporarily unavailable the real world blind super resolution , right?

hellloxiaotian commented 2 years ago

Yes, you can modify the test_dataset class to make it available.