search

zwx8981 / DBCNN-PyTorch

An experimental Pytorch implementation of Blind Image Quality Assessment Using A Deep Bilinear Convolutional Neural Network
MIT License
215 stars 32 forks source link

Using simple_demo.py #29

Closed kangzhao2 closed 2 years ago

kangzhao2 commented 2 years ago

Hi author:

I want to use simple_demo.py to test jpg quality. I set:

scnn_root = 'pretrained_scnn/scnn.pkl'

// ckpt = "your path of the checkpoint file" image_name = "4.jpg" // checkpoint = torch.load(ckpt) // model.load_state_dict(checkpoint)

I comment the code of checkpoint, since it has been set in scnn_root, am I right?

“4.jpg” is my test image, and I get the output: Prediction = -0.0054. I think the prediction should be [0, 1], so whats wrong with my codes?

Kang

xiongxiongtiao commented 2 years ago

No, you should first train the model on an IQA database, and set the ckpt to the path of the trained checkpoint.

The range of the output depends on the range of MOS/DMOS of the IQA on which your model is trained.

Without trained on any IQA dataset, the weights of the model is just a meaningless random intilizations.

发送自 Windows 10 版邮件应用

发件人: Kang 发送时间: 2022年1月5日 19:40 收件人: zwx8981/DBCNN-PyTorch 抄送: Subscribed 主题: [zwx8981/DBCNN-PyTorch] Using simple_demo.py (Issue #29)

Hi: I want to use simple_demo.py to test jpg quality. I set: scnn_root = 'pretrained_scnn/scnn.pkl' ckpt = "your path of the checkpoint file" image_name = "4.jpg" checkpoint = torch.load(ckpt) model.load_state_dict(checkpoint) I comment the code of checkpoint, since it has been set in scnn_root, am I right? “4.jpg” is my test image, and I get the output: Prediction = -0.0054. I think the prediction should be [0, 1], so whats wrong with my codes? Kang — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: @.***>