About the supervised problem.

[mtroym]

🔥 You used the GT labels from the training set, why you could claim yourself unsupervised segmented the primary object? BTW, maybe I have some mis-understanding about the DAVIS unsupervised TEASER challenge. Did it mean when inferencing, the given data is only video sequence. And we could use the GT labels as training label? (I think it is not the so-called unsupervised learning.)

[wenguanwang]

@MTonyM

See Sec. 2.2 in "A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation".

In video segmentation area, the unsupervised/semi-supervised/interactive segmentation are depended on whether the labels are given during testing. Note that it is different from unsupervised/semi-supervised learning strategy.

[mtroym]

Thank you for your reply during great new year. Same as what I thought. But there is some confuse about how to do unsupervised VOC without label during training? How? Thanks!

[wenguanwang]

@MTonyM

As I stated before, the unsupervised segmentation only means we do not use the labels or mutual interactions during testing. You can use a supervised or unsupervised learning technique during training. I think most video object segmentation work such as "A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation" have already clarified this point very clear.

[mtroym]

thanks! @wenguanwang