Question about the Beta parameter of Foggy Cityscapes

kinredon / umt

A Pytorch Implementation of Unbiased Mean Teacher for Cross-domain Object Detection (CVPR 2021)

MIT License

85 stars 7 forks source link

Question about the Beta parameter of Foggy Cityscapes #15

Closed lch1999 closed 1 year ago

lch1999 commented 1 year ago

The Foggy Cityscapes have three kinds of images, including beta=0.005, 0.1 and 0.2. Could you tell me which kind of images are used in this paper? Thank you very much!

kinredon commented 1 year ago

The worst foggy, i.e., 0.2

lch1999 commented 1 year ago

Thank you!