Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Copy Feature v.09 #221

Open
MaxPRon opened this issue Mar 15, 2024 · 11 comments
Open

Copy Feature v.09 #221

MaxPRon opened this issue Mar 15, 2024 · 11 comments
Assignees

Comments

@MaxPRon
Copy link

MaxPRon commented Mar 15, 2024

Hi,

I'm using the Xtreme Version 0.9 locally to label a lidar-(fusion) dataset.

I wanted to ask if there is a copy feature available ?

It would already be super helpful just be able to simply copy the labels into the next frame. If seen it's possible in the cloud version. In #100 (March, 2023) it was mentioned that tracking is not planned to be included in the community version.

Does this still apply ? Again a simple copy feature without any compensation etc. would already be nice.

Maybe there is also a workaround.

Thanks a lot!

@MaxPRon
Copy link
Author

MaxPRon commented Mar 21, 2024

Hi :) @jotamotk Is there any update on that issue ?

@jotamotk
Copy link
Collaborator

jotamotk commented Mar 22, 2024

Hi @MaxPRon ,
Sorry to reply to you late. I was otherwise engaged.

Copy features are currently unavailable in Xtreme1 but are planned in the future. Our new roadmap will include integrating and open-sourcing almost all of the BasicAI cloud toolsets into Xtreme1.

Due to limited resources, our next priority will be integrating extensive toolsets like text, LiDAR Segmentation, and LLM into Xtreme1 instead of advanced features like copy. So it will probably be late. If you want this feature shortly, any PRs will be appreciated.

Best,
James

@jaggerwang jaggerwang assigned guhaomine and unassigned jotamotk Mar 22, 2024
@MaxPRon
Copy link
Author

MaxPRon commented Mar 22, 2024

Hi @jotamotk ,

thanks a lot for the update.

Regarding the feature, I could give it a try but we super helpful if you would be able to give some pointers where in the code I should get started.

I would aim for a simple copy feature with the following funcitonality:

  • Button/Check mark to select "copy mode"
  • Copy/Merge the existing label (file) with the to/with the one from the next frame

Best,

Max

@jotamotk
Copy link
Collaborator

Sure, @guhaomine will give you some pointers to start

@guhaomine
Copy link
Collaborator

guhaomine commented Mar 27, 2024

@MaxPRon image
In the point cloud continuous frame data set annotation, we implemented the function of copying the results to the next frame.
In the code, there is a method called "track" that you may need. The file path is
'frontend\pc-tool\src\packages\pc-editor\common\DataManager.ts'
image

@MaxPRon
Copy link
Author

MaxPRon commented Mar 28, 2024

@guhaomine Thanks for the pointer. Another quick question. The screenshot you're showing is from the cloud or community version ?

The code snipped you mentioned you used is currently not used, correct ?

@guhaomine
Copy link
Collaborator

@MaxPRon No, the community version has this feature, you need to upload a continuous frame data set. The timeline will be displayed
image

@chanyoung1998
Copy link
Contributor

chanyoung1998 commented Apr 17, 2024

@MaxPRon is there any update? our team is planning to implement that object tracking function for our own. If you have any update or progress about it, let me know!

@MaxPRon
Copy link
Author

MaxPRon commented Apr 17, 2024

@chanyoung1998 Not really, the hint @guhaomine gave was actually what I was looking for. Although l would still be interested in an object tracking/interpolation feature.

@guhaomine Also thanks a lot for the hint :)

@chanyoung1998
Copy link
Contributor

@guhaomine , if you happen to know model request format for Tracking Object ? I'd like to know http call request format for tracking model.

@jaggerwang
Copy link
Contributor

@chanyoung1998 Right now, only object detection models can be customized, the object tracking model is a little complicated, as it need to call the object detection model first, it will be simplified and allow customize in the future.

You can find the related docs here.
image

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants