I'm trying to find a way to use the Nutrionix API to work with my model. I don't have the entire database, which is problematic, but perhaps running a script will give me enough data for the model to learn. The worst case is that I have to get a license for the DB with money.
I will have to calculate the calorie density of the food and use that as the metric. I will try with calories/gram, however, this might be too simplistic in the vast data of food. One simple example of a food with a lot of density with an average calorie/gram ratio is popcorn. This is due to the large amount of air in popcorn. This causes an issue in using the calories/gram metric as air doesn't have weight. Perhaps another metric can be used: calories/serving. However, this causes the issue of the serving size being chosen by the manufacturer. Many foods have very small serving sizes, leading the consumer to think that a food has 200 calories, while in reality it has 400 as it has two serving sizes. But a potentially perfect metric might be calorie/cup. Cup is a volume metric, and is perfect for my use case. Work needs to be done in this area.
Top comments (0)