search

bair-climate-initiative / scale-mae

Make your models invariant to changes in scale.
https://ai-climate.berkeley.edu/scale-mae-website/
Other
121 stars 13 forks source link

How to use pre-trained models for semantic segmentation tasks？ #12

Open Magic-lem opened 8 months ago

Magic-lem commented 8 months ago

Thank you for your great work. In your paper, you used UpperNet to implement image semantic segmentation. Can you provide the relevant code?

Thank you very much.

Shanci-Li commented 8 months ago

Hello!

Do you have any updates about the semantic segmentation implementation? I am interested as well :)

Magic-lem commented 8 months ago

I implemented it using the code in this repository: https://github.com/CSAILVision/semantic-segmentation-pytorch

RitwikGupta commented 8 months ago

We ran these experiments using mmseg!

Magic-lem commented 8 months ago

Thank you for your reply. Furthermore, what layers of output do you use to input into UpperNet?

Mayolov commented 3 months ago

@Magic-lem Can you share the code you used to get the same segmentation results?