Iterate / cycle through objects in Standard Mode

[aguadopd]

My actions before raising this issue

• Read/searched the docs
• Searched past issues

It would be helpful to be able to go object by object editing annotations, while on Standard mode.

Expected Behaviour

• Annotators can iterate through all objects in the frame, or a filtered subset of them, editing them 1 by 1 and minimizing the risk of skipping one.
  • They can edit back and front of an object, just by changing between frames and having previously disabled the Reset zoom option.
    • This implies either having a method for linking objects (is there something like this already?) or just changing frame without moving the scene, since the object should be nearby in the surrounding frames ---in the same position in our case---. We already can do the latter, if disabling Reset zoom.
  • They can also delete objects.
  • They can also create new objects, either by drawing new shapes or by splitting already existing ones ( Cut one shape into two parts within a frame #690 ). These new objects go to the end of the list so this doesn't affect the workflow by introducing stuff that won't be visited.

Current Behaviour

• You cannot focus on objects in a sequential fashion while at Standard Mode.
• The iteration behaviour is already present in the Attribute Annotation Mode, where you can inspect objects one by one.
  • But you can't easily see the same object in some other frame, since when you change frame, CVAT focuses on the first object of the current frame.
• Currently, the only shape modification you can do in AAM is to move points/vertices. If you try to edit a shape, you suddenly get focused on the first object of that frame, making it unusable. I'll detail that in another bug, but maybe it's expected behavior since this is for editing Attributes.

Possible Solution

• I guess one could use TAB and Shift+TAB to cycle through the objects while at the Standard Mode, just as it currrently works in Attribute Annotation Mode.

Context

We want to use CVAT to edit/correct the automated annotations made by a CV segmentation program. The corrected images are used for testing and further developing the segmenters.

• Our images have different views (currently only 2, front and back) of the same objects, and this means that, besides well segmented, we also want to have the same label on all of the views that belong to the same object.
• We are currently embedding the views as frames into CVAT, uploading 2 images per task; they make a job of 2 frames.
• There are many (>100) objects per image. These are small so users have to work in a zoomed-in view, moving around the scene/canvas.

Your Environment

• Git hash commit (git log -1): 18f6b2f
• Docker version docker version (e.g. Docker 17.0.05): 19.03.9
• Are you using Docker Swarm or Kubernetes? No
• Operating System and version (e.g. Linux, Windows, MacOS): Ubuntu 16.04

Next steps

You may join our Gitter channel for community support.