search

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

What if I didn't use BL30K? #14

Closed longmalongma closed 3 years ago

longmalongma commented 3 years ago

I did not use your pretrainined model and started training your network again. My pre-training only used static pictures instead of BL30K. After the completion of the training, the result of the test on Davis 2017 data set was only 70.6%(J F mean), is this result normal?What would the normal result be if we didn't use BL30K?

image

hkchengrex commented 3 years ago

You should get a J&F of 83.1. This is provided in a table of the readme.

longmalongma commented 3 years ago

You should get a J&F of 83.1. This is provided in a table of the readme.

OK, I will retrain again.