Image recognition & suggested auto-crop

[bredowmax]

What

• This works best for products that have been scanned in the shelf rather than in the hand of a consumer. Scanning products in the shelf (next to the corresponding price tag with optionally a barcode) is the fastest way to lift as many products onto the platform as possible and improves scanning experience for users as we can already pull some basic information about the product before the user adds nutrition table & list of ingredients.

Describe the solution you'd like
I'd like the primary image to try to crop into a product (with sufficient space around)

Describe alternatives you've considered
It's possible doing this by hand. Once you did this task 500 times, it feels like a machine would be able to do it because of its repetitive nature.

Additional context
In a distant future, an OFF scan party could look like this:
https://www.youtube.com/watch?v=SXEaCZ4KzL8
https://www.youtube.com/watch?v=KRJV1SPYpIE

This product

Would look like this

[teolemon]

0099482478049

[github-actions]

Stale issue message