search

Zheng-MJ / SMFANet

[ECCV 2024] SMFANet: A Lightweight Self-Modulation Feature Aggregation Network for Efficient Image Super-Resolution
Apache License 2.0
82 stars 5 forks source link

model weight issue #2

Open liuxiaoya396 opened 2 months ago

liuxiaoya396 commented 2 months ago

Hello, there seems to be an issue with the weight of the x4 model. The prediction results have not improved, only the resolution has increased

Zheng-MJ commented 2 months ago

Our method is based on the bicubic downsampling of the input, which may exhibit a domain gap for real-world degradation inputs. If you are interested in attempting super-resolution for real-world scenarios, you may try our another work (x4 Real-world): https://github.com/sunny2109/SAFMN