add parameter to "union" all boxes at each timepoint

[keighrim]

New Feature Summary

Instead of having all the individual localization boxes, we'd like to have a option to "union" of all boxes into a one big box.

implementation plan

• add a parameter mergeBoxes or unionBoxes.
• after EAST localization is run and all boxes are added to a new view, if parameter is true, create a new view and add "union" boxes with boxType=text
• for creating union boxes, a simple implementation will be using top-most, bottom-most, left-most, right-most coordinates of all boxes to create one big box.

Related

This implementation will be used for "reading-order" selection in downstream OCR apps, such as clamsproject/app-tesseractocr-wrapper#18 .

Alternatives

No response

Additional context

No response

[keighrim]

Just wonder if it makes sense we have an independent app for doing this "merge" operation, based on TimePoint information and matching boxType values. Having an separate app will enable us to use different text localization app in the future.

[MrSqually]

this makes sense to me. Currently, the merge parameter concatenates an entire new view + set of bounding boxes to the mmif anyway. moving this to a separate app makes more sense, cleans up the code, and reinforces the idea that any given View is a specific tool's output. im quite busy today, but I could pretty much drag+drop my current implementation to a new clamsapp so it shouldn't be hard to implement.

[keighrim]

Let's do that then!