Regarding the upsampling layer

aurora95 / Keras-FCN

Keras-tensorflow implementation of Fully Convolutional Networks for Semantic Segmentation（Unfinished）

MIT License

650 stars 268 forks source link

Regarding the upsampling layer #83

Open netw0rkf10w opened 5 years ago

netw0rkf10w commented 5 years ago

Hi,

Thanks a lot for the implementation!

Regarding upsampling, you used a custom BilinearUpSampling2D layer, which I guess cannot be learned (sorry if I'm mistaken).

I believe a better solution is to use the built-in Conv2DTranspose Keras layer for upconvolution (c.f. section 3.3 of the FCN paper).

Best regards.

netw0rkf10w commented 5 years ago

I have checked the code again and there seems to have a more serious problem: there is no skip connections at all between coarse and fine output layers :(

NickChang97 commented 4 years ago

Yes I found it also. There is no skip connections, a little pity. But the code's structure is nice