Thanks for reaching out and elaborating.
We are constantly working on improving our Image Recognition models with monthly updates enhancing speed, accuracy in food detection and overall food matching against the respective local country dataset.
There might instances where complex food photos might not return all ingredients and serving amounts at the expected level.
Hence, if the Image Recognition API does not return fully satisfactory results we recommend:
- Either having the image being scanned again
OR
- Having a user interface where the user can see all the individual foods and serving amounts returned being displayed and editable i.e. the user can chose to change either the serving amount and/or swap out the food itself against another food from that local country database. We will have such a UI available for demo in our own mobile app in the coming weeks for comparison.