Augmentation on multilayered images

[sudo rm -rf --no-preserve-root /]

User Feedback:
Keep in mind, some of these augmentation tasks will need to be applied to the ground truth image labels as well (e.g. masks). Mainly the ones that change the scale, occlude, or orientation of the image, not the ones the deal with the pixel intensity or color.

-> some augmentation options must be applied to the ground truth labels as well

[sudo rm -rf --no-preserve-root /]

The user can upload multilayered images to the projects. A multilayered image is an image that consists of multiple layers. Think of it as several images combined into one image.

Why should the user need to work with multilayered images?
You might need to work with multilayered images when you need several references of the same image to achieve higher augmentation and annotation accuracy. This is applicable for e.g. augmenting and annotating aerial and satellite imagery. For instance, you can upload an image consisting of 3 layers: SAR (synthetic aperture radar imagery), IR (infrared satellite imagery), and RGB.

How does it work?
When you work on one layer, the changes will be applied to the other layers as well. So if you add augmentations (or later annotations or comments) on one layer, the augmentations (or later annotations and comments) will also appear on the other layers.

[GiaoNguyen]

Total agree that, the augmentation should effect for all layers. We will add this futures.

@pcaversaccio, we should think about how to upload data with multilayers? How to find the relation between the different layers of an image. cc @ttattl

[sudo rm -rf --no-preserve-root /]

Currently on hold. For further details see the roadmap here.