microsoft / unilm

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
https://aka.ms/GeneralAI
MIT License
20.19k stars 2.55k forks source link

about input image size #1054

Open menghuaa opened 1 year ago

menghuaa commented 1 year ago

Describe Model I am using beit base 224224: Hello, Can the beit model only extract features from fixed size images? When I extract features from images with a resolution of 1920 1080, it will report an error "Input image heigth(512) doesn't match model (224)". How can I modify it so that it can extract features from images of any size?

addf400 commented 1 year ago

interpolate the position embedding, more details can be found at here and here