I'm using ODK Collect to collect images that will be used by a computer vision algorithm. I would like to be able to present some sort of bounding box (like what banks use on check photos) in the view of the camera itself or, even better, to be able to have the view automatically save an image and close when it recognizes that the image is in the correct orientation (like what many qr code applications do). It is my understanding that it is not possible to modify the image question type to do this in ODK 1. Is this kind of extensibility possible with ODK 2? I would also be happy to submit a patch that implemented an image bounding box as a new question type.
p.s. I wasn't sure if this was more appropriate for Support, but I thought I might as well start here.
ODK Collect does have a
align image type which uses an external app to display an overlay: https://github.com/opendatakit/collect/issues/775. It's very clunky and I personally think it would make most sense to replace with something better but if you have a time-sensitive application, you may be able to leverage it now.
Longer-term, I think we can and should add this functionality natively. The first step would be to figure out how someone would specify the bounds at the form level. For example, it could be a set of attributes to define corners of a bounding box or it could be a single attribute that defines something like a string of arbitrary x,y coordinates so that any polygon can be overlaid.
Okay that sounds enticing. I'll look into options and report back soon!