GuHuangAI / DiffusionEdge

Code for AAAI 2024 paper: "DiffusionEdge: Diffusion Probabilistic Model for Crisp Edge Detection"
Apache License 2.0
236 stars 25 forks source link

Using images with just 1 channel instead of RGB #18

Closed f-hader closed 8 months ago

f-hader commented 8 months ago

Hi, I am interested in trying out your model for analyzing some measurement data that has only one channel. How can I customize your model and training code to work for single channel inputs? I don't think it's a good option to just store the same data on three channels...

GuHuangAI commented 8 months ago

Hello, you can use the modified backbone for single-channel image. Please refer to the single channel Swin-transformer here.

f-hader commented 8 months ago

Thank you very much, I was able to modify the network and training code for 1D and will start training now.