after training a denoising network from dnCNNLayers, how to use it in denoisingNetwork and denoiseImage?

Question

Ming-Jer Tsai 2020-8-31

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/587000-after-training-a-denoising-network-from-dncnnlayers-how-to-use-it-in-denoisingnetwork-and-denoiseim

评论： Ming-Jer Tsai 2020-9-16

I name my denoising network as dncnn_xfer, which is trained starting from dnCNNLayers. Since denoiseNetwork('dncnn') only accepts 'dncnn' I rename dncnn_xfer as dncnn, dncnn=dncnn_xfer. But net=denoiseNetwork ('dncnn') seems to always pick up the pretrained DnCNN existing in MATLIB instead of dncnn_fer.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Madhav Thakker 2020-9-16

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/587000-after-training-a-denoising-network-from-dncnnlayers-how-to-use-it-in-denoisingnetwork-and-denoiseim#answer_495325

Hi Ming-Jer,

I understand that you want to use your custom denoising network. I assume you have trained the network sucessfully using the procedure. After training the network using trainnetwork, it returns a trained network in the output argument. There is no need to initialize a network using denoisingNetwork. denoiseImage can accepts as input the trained network and returns a denoised image.

Hope this helps.