How to skip the pre-training of static images and BK30L and go directly to VOS training?

hkchengrex / Mask-Propagation

[CVPR 2021] MiVOS - Mask Propagation module. Reproduced STM (and better) with training code :star2:. Semi-supervised video object segmentation evaluation.

https://hkchengrex.github.io/MiVOS/

MIT License

127 stars 22 forks source link

How to skip the pre-training of static images and BK30L and go directly to VOS training? #12

Closed longmalongma closed 3 years ago

longmalongma commented 3 years ago

How to skip the pre-training of static images and BK30L and go directly to VOS training? What changes need to be made in the code and command line?

hkchengrex commented 3 years ago

Yes. Just start stage 2 without loading a pretrained model (--load_network).