search

nhshin-mcl / QCN

[CVPR 2024] Blind Image Quality Assessment Based On Geometric Order Learning
MIT License
13 stars 0 forks source link

How to use a trained model to perform inference on other images and get a score? #2

Open AngryPigPig opened 2 months ago

AngryPigPig commented 2 months ago

Hello! First of all, I'd like to say that your paper proposes a very useful and feasible method for no-reference image quality assessment, and I believe that this method can empower many industries in the future. After reading your paper, I tried training the model myself and obtained a model with good indicators, SRCC_Epoch_99_SRCC_0.9225_PCC_0.9251_MAE_6.2787.pth. Now, I want to use my model to perform inference tests on other images, but I get an error when calling the model.eval() method after torch.load: AttributeError: 'dict' object has no attribute 'eval'. I would like to ask if there was an error in my training or if I am calling the method incorrectly?

Looking forward to your reply, thank you!

nhshin-mcl commented 2 months ago

Hello! Sorry for the late reply. I think there is a problem with calling the method. Please check the parameter loading part in the codes. Thank you for your interest.