image_manual questions

A couple of questions/observations about the image annotator.

  1. It's currently possible to draw a shape and drag it off the canvas. Once released, if the label is off canvas then it is no longer selectable and one must clear all current work in the tool.
  2. If you do export it, the shape reports negative coordinates. I would think that this should probably 0 out to top/left and max width/height for bottom/right?"type":"rect","points":[[-82.3,-33.3],[-82.3,243.5],[350.6,243.5],[350.6,-33.3]]}]. I'm currently checking for this in post-proccess, and maybe that's your intent.
  3. Feature request for an event listener (double click?) on the label or shape itself that would expand a shape to the full width/height of an image. Use case: I am often finding that as I annotate a set of images that a label may apply to regions within but often can just apply to the whole image as well and dragging for full width/height can be 'slow' on bigger/tall images.

Welcome to the forum @whicks1 and apologies for the late reply!

Thank you for your valuable suggestions; we will certainly include them in our improvements backlog. Regarding the issue of negative coordinates, a possible solution could be to, actually, restrict the ability to drag the shape off the canvas. The usefulness of it is definitely debatable.