specifications for interactive learning and segmentation app about canvas-sandbox HOT 4 OPEN

Ok great! Is it possible to overlay multiple images with add_layout_image so that the image uploads can be faster (we can just send say a png which is transparent where the class isn't)?

from canvas-sandbox.

Yes, add_layout_image can be used to add several images, possibly with transparency. Maybe there is room for a small optimization here :-).

from canvas-sandbox.

Ah it's true, currently there's no way to just add / remove specific images from the graph's figure? We just have to upload them all each time? Say I display 7 images and I want to remove 1, I have to upload 6? Or I guess the browser might deal with this by chaching images...

from canvas-sandbox.

Update: https://github.com/scikit-image/scikit-image/blob/50d107fb042abb01dfc4f63225c463409e3f6112/doc/examples/segmentation/plot_trainable_segmentation.py can be used to define features and compute the segmentation.

The Dash components to select features can mirror the arguments of the trainable_segmentation function, that is there can be a checklist for intensity, edge and texture, a rangeslider for selecting the sigma range, and an input for the downsample parameter.

from canvas-sandbox.