Transfer learning using UNet

Question

1 个投票

I'm currently working through the "Deep Learning for Semantic Sementation" https://ch.mathworks.com/help/vision/ug/semantic-segmentation-using-deep-learning.html workbook to try and augment my unet segmentation algorithm. Is there anyway of conducting transfer learning with unetlayers? It says in the workbook it's possible with unet and I've successfully done it with the other segmentation architectures suggested (fcn and segnet). Any help appreciated!

5 个评论
显示 3更早的评论隐藏 3更早的评论

Matt J 2022-10-20

So the problem is finding a pretrained U-Net, not the actual transfer learning?

Lilly 2022-10-20

编辑：Lilly 2022-10-20

No, that's not the problem. If you go onto the "Deep Learning for Semantic Segmentation" documentation I've linked in the 1st post. The documentation says:

"Create the Network

Use the deeplabv3plusLayers function to create a DeepLab v3+ network based on ResNet-18. Choosing the best network for your application requires empirical analysis and is another level of hyperparameter tuning. For example, you can experiment with different base networks such as ResNet-50 or MobileNet v2, or you can try other semantic segmentation network architectures such as SegNet, fully convolutional networks (FCN), or U-Net."

I can use the deeplabv3pluslayers function to make a transfer learning network bacsed on ResNet-18 and segnetlayers with vgg16 but I can't do that using the same syntax with the unetlayers function with a pretrained model. I'm just wondering whether there is another way to use transfer learning to make a unet that is similar to unetlayers but allows transfer learning

请先登录，再进行评论。

请先登录，再回答此问题。

Follow Question