Allergens sometimes are written in other part of packaging than ingredients
So my question is if information about allergens and possible allergens should be included in ingredients photo or not?
Which variant will be correct?
This one:
Or this one:
?
When submitting pictures, I take one general picture of the whole product per side. In your example, it seems your product has two sides, so I would take two pictures. I would add those pictures to the mobile app.
Then, I would reuse those pictures to fill the tiles “Product picture”, “Ingredients picture”, “Nutrition”, “Packaging”.
Each time, I would crop one of my two pictures to the relevant section.
For the ingredients, I would crop it as in your first variant, with only the ingredients.
If there would be a tile for “Allergens”, I would put your second variant in it, but it is not the case. Reusers can look at my general picture to get the allergens section.
But it is important in my opinion to include in general pictures all the product information, including one that do not fit in the four tiles. This enables verifiability.
Hi, this is exactly what I do after I have found, that if I crope existing pictures they are not stored as separate and instead of that it just records relevant coordinates of the original picture. So I take general pictures of some product from all sides and after that just choose and crop them to save storage on OFF servers.
So I agree and act already with this practice.
I had same thoughts but wasn’t sure, because in fact allergen info is ingredient info as well, so I wanted to check it
I understand the ingredients as the list of things I would need listed if I was going to prepare the recipe myself. When I prepare a recipe, I do not need the allergens list to make the recipe. Only the ingredients. The allergens is useful if I am going to eat it, but not for preparing it.
That’s only my understanding, other people may think differently.
I think @Canard164 interpretation is the right one.
More technically ingredients image is passed to OCR and Open Food Facts server tries to get the ingredients list from it, so the less noise, the better.
So @Canard164 @alex in fact, according to your opinions, these images should be re-cropped in the next way, right?
Product barcode: 0013764027053
Product barcode: 3046920028004
I think so, yes.
Without losing any information though, so as already mentioned, pictures of the whole product and packaging should also be present to allow verification of everything, like allergens.
@
@Libra I don’t tend to be so precise (but for machine learning data annotation).
Personally I would keep the allergen part if they are just below. But I would remove the “contains” from the extracted ingredients (normally the server does it automatically for some languages). But really it’s a matter of personal way of doing.
The general philosophy of Open Food Facts, as a crowdsourcing project, is to grab data, as is (best effort from users from beginners to experts), and then process it to get structured data. So a bit of noise in raw data is expected.
Of course I understand that a bit of noise is totally OK. I don’t expect it will be 100% correct and same. The key problem is at least for now that it is not “a bit noise”. I don’t see any data consistency almost at all. Everyone edits according to their own vision only and it makes me cry, because it makes the data too unstructured and almost useless without manual sorting and working with this data. It limits automatically any massive data analysis, and it sucks.
I’m also a contributor to OSM and MB. They are crowdsourcing projects as well, but their data is much more useful and it is much easier to use it, because their data is much more structured and unified. They have guidelines how things should be added to their projects and contributors should behave according to these guidelines. Even they are not always strict, it still helps to make data much more unified. Yes, not always new users will know and read all of these guidelines, but in cases when they behave incorrectly, there are experienced users that can help fix and explain how they should to do similar things in future.
This is one more thing, that I hate about OFF right now. There is no literally any solution for communicating between project contributors. Slack is global, it is not its goal to be used for personal discussing or for discussions about 1 specific product only. If someone vandalizes products or make random incorrect edits on products I added I will not even know about that, because there is no any mechanism for watching products. There is no any direct way to speak with a specific editor in OFF.
Contribution guidelines are hidden deeply in Wiki, don’t have details, while guidelines in the mobile app are unclear.
In addition, OFF editors (the website and the app) are uncomfortable to use. The app even adds English localization to all my products I add, even if they don’t have it…
Storing of brands, stores and origin places is unstructured at all.
So yes, as I said many times, at least for me it looks like the project doesn’t have any data consistency policy and it makes that hard to re-use its data. It is not possible to work with data, when literally for every product you as re-user need to understand how data for this specific product is stored, and you need to do this process for every product separately. It is not library or store, in fact it is just data dump, that you need to clean and filter it firstly to be able to do something with that.
Except for quantity, nutritional data and images, all other data is unstructured and inconsistent unfortunately.
I really like the project, its idea and its mission, but from all crowdsourcing projects I know, unfortunately, I need to say this project is the weakest, if we speak about potential data re-using.
