[Question] Zooming/panning/annotating contextual images in sync with main image

[jonvanveen]

Hello,

I'm working with tracking fairly small (close to point-source) objects across 2D frames.

• I would like to be able to have contextual images zoom and pan in sync with the zooming/panning I do in the main image, so that the same zoomed-in/panned view is maintained across the main image and the contextual images.
• I would also like to have annotations sync to the contextual image, so that an annotation I make on the main image is annotated simultaneously on the corresponding object on all the contextual images. This would be helpful where our context images are a couple of temporally previous and next frames, to speed up the annotation process for many objects. I don't think this exists given that contextual images are entirely separate files, instead of being able to use the previous/next frames in the main image sequence, but figured I would ask.

Is there any functionality for this in CVAT? If not are there any plans for it? I couldn't find anything relating to this in the docs or the source code.

Thank you!

[bsekachev]

Hello

I believe that would be a nice feature.
Unfortunately now there are not anything like this.

[SergeyMikhalcov]

Hello,

My colleague made contextual image zooming not in sync with main image, but it may good point for starting. Link for issue and link for pull request

Hello

I believe that would be a nice feature. Unfortunately now there are not anything like this.