🔴 Confirm dish name after applying image recognition.

When using our recognition endpoints the most probable classes are returned concerning dish, drinks, sauces or ingredients (depending on the type).

With this endpoint, the user has the option to confirm which one of the returned food classes is the correct one.

By default, the source is set to 'logmeal' and an ID corresponding to an existent dish is expected. The 'source' can also take the value 'other' for specifying a free-text feedback for any dish that is not included in LogMeal's API. Do note that if you confirm a dish with a free-text feedback (source 'other') it won't be taken into consideration when doing any calculation that is based on the dish (e.g. generating ingredients or nutritional information).

If a food segmentation endpoint has been used then you have to provide a food item identifier for each of the confirmed dishes. This implies adding the extra parameter 'food_item_position'. This parameter consists of a list of elements, one per each confirmed dish, and it must include an integer identifying the segmented region it corresponds to (see 'food_item_position' in returned json for /segmentation endpoints). If you desire to confirm a dish that does not belong to any of the detected regions then a string has to be provided that will act as an unique identifier for that extra confirmed dish.

Important notes about dish IDs:

• If the image is a default API intake, the dish IDs you can confirm must belong to LogMeal’s global dataset. You can check the full list of available dishes in the endpoint /dataset/dishes.
• If the image is a kiosk/tray intake, then the dish IDs must match the dishes actually offered that day and shift in the restaurant. These valid dish IDs can be retrieved from the endpoint /kiosk/external/dishes.

To know if an image is of type kiosk, you can check the endpoints [GET] /intake or /getIntakesList. If the field is_kiosk_image is true, then you must use the kiosk dish IDs instead of the global dataset.

Example payload

Suppose an image has already detected dishes with IDs 2284, 2270, and 2290. If you want to confirm those three and also add a new dish with ID 2267, you need to send all the confirmed IDs and their positions, plus the new one:

{
  "imageId": 1982196,
  "confirmedClass": [ 2284, 2270, 2299, 2267 ],
  "source": [ "logmeal", "logmeal", "logmeal", "logmeal" ],
  "food_item_position": [ 1, 2, 3, 4 ]
}

Each list must include both the existing confirmed dishes and any new ones you are adding.