Allergy Pal

Problem

Between 1.5-5% of Americans have experienced anaphylactic shock in their lifetime. Serious food allergies are a health condition that can affect anyone, regardless of their level of preparedness and access to medical services. To help prevent life-threatening allergic reactions, we created Allergy Pal

Functionality

Allergy Pal is an easy-to-use web application that allows users to stream the webcam on their device or share an image from storage to find the ingredients from a displayed food item in the frame. Allergy Pal will then produce a list of common ingredients for that dish, as well as allergens that are occasionally included. These ingredients will be cross-referenced with a user-inputted list of allergens and warn if any matches are detected.

Processes

Allergy Pal functions off of a Tensorflow machine learning model trained on over 100,000 images of 150+ different dishes. The detected dish is then passed through OpenAI's GPT-3.5 Turbo API to produce a list of ingredients shared by hundreds of recipes online.

Scalability

We wish to expand the machine learning dataset for Allergy Pal to include at least 250 loaded food items, as well store user login informations and allergens in a database, to prevent users from logging allergies per each use.

Built With

• css
• html
• javascript
• node.js
• openai
• python
• tensorflow