[Feature Request] Rapid online assessment of Augmented Image quality

[AlgoRyan]

Given that selecting high-quality image targets is so important for creating a stable AR experience it makes sense to enable developers to validate the best image target as quickly as possible.

The arcoreimg tool is already a great step in the right direction here, but running the script on a local machine is still a pain point when prototyping image targets. It's also a barrier to entry for new teams considering using ARCore, since it's difficult for them to get a sense for what makes a good target.

Could we please add some functionality to the reference documentation page to enable a drag-and-drop image validation flow? Something as basic as outputting the score for a given image would enable designers to test prototype images without developer assistance.

Logical extensions could include:

• Displaying the recognised feature points, to convey which parts of the image need work
• Providing broad advice about issues with the uploaded image (e.g. unnecessarily high resolution, compression issues)
• Breaking down the causes of a low score (i.e low feature count vs repetitive features)

I'm imagining that this is on the roadmap already, but felt it was worth tracking in a public forum. Let us know if you have something like this planned!

[inio]

Thanks for the suggestion. I've noted it in our internal tracking system.

[Kjelll]

Also just a basic online assessment of

drop image -> image score is X

would be really helpful already