search

yoxu515 / aot-benchmark

An efficient modular implementation of Associating Objects with Transformers for Video Object Segmentation in PyTorch
BSD 3-Clause "New" or "Revised" License
602 stars 108 forks source link

The mask of first frame #17

Closed Carter-1998 closed 2 years ago

Carter-1998 commented 2 years ago

I would like to ask if the target to be split does not appear in the mask of the first frame, can it still be split smoothly in subsequent splits?

z-x-yang commented 2 years ago

Sorry, I can't understand what you point to.

Carter-1998 commented 2 years ago

I apologise for not being clear. I meant if the target of interest does not appear in the first frame, can it still be segmented in subsequent frames？

z-x-yang commented 2 years ago

Yes, the framework can do this well. The cases you talked about can be found in many videos on YouTube-VOS.

Carter-1998 commented 2 years ago

Thanks a lot.

---Original--- From: "Zongxin @.> Date: Thu, Apr 14, 2022 22:41 PM To: @.>; Cc: @.**@.>; Subject: Re: [yoxu515/aot-benchmark] The mask of first frame (Issue #17)

Yes, the framework can do this well. The cases you talked about can be found in many videos on YouTube-VOS.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

9p15p commented 2 years ago

Hi, @Carter-1998 @z-x-yang

It seems that Semi-supervised VOS needs the reference mask in the first frame. If we do not provide it, How can the model know which object to segment?