4 ms·
Great start for a product. I would imagine everything that's currently in your food database had to be manually chosen due to the availability of training data.
by bmcooley 10y ago
Great start for a product. I would imagine everything that's currently in your food database had to be manually chosen due to the availability of training data. It doesn't handle very many branded products, but that could be really valuable in keeping users engaged and using your application daily, so I would suggest adding barcode scanning (currently building an application with the Nutritionix API, its got a great dataset). You could even ask your users to take a picture of their barcode entered food so you can start learning on a much wider variety of products. My two cents.
- m_ke 10y agoHey, thanks for the feedback. All of our nutrition data comes from the USDA right now. We don't recognize everything that’s in there yet but we map our predictions onto some of the “nodes” and let people refine the predictions to a more specific item. We don’t recognize packaged products yet but plan on doing it once we have enough data. Barcode scanning is almost done and should make it into the app soon.