Inspiration
The cosmetics industry is worth over $50 billion USD, and just about every person in the world buys some cosmetic products, such as deodorant or makeup. However, between outdated regulations and uninformative ingredients list, being an informed consumer of cosmetics is near impossible! Since many skin care products (e.g., moisturizers, oils) are intended to be absorbed into the body, people should know about potentially harmful substances within the products they are buying.
What it does
We built a phone app that can provide near instant information to people shopping for cosmetics. We allow them to simply take a picture of a list of ingredients and display the purpose of those ingredients, as well as flag any that are potential carcinogens identified by the WHO’s International Agency for Research on Cancer.
How we built it
We built the app using React and Flask. We used Wikipedia’s API to find information on specific chemicals. The World Health Organization’s International Agency for Research on Cancer website was also used to determine whether an ingredient is harmful based on its classification. The Tesseract Optical Character Recognition library was used to recognize the text within an image.
Challenges we ran into
Initially, we searched the WHO's database by running through the list in order to match it to the compound name. However, it turns out that many substances have multiple ways of writing their name. For example, Salicylate can also be written as Salicylic acid. To solve this problem, we had to use a lot of Natural Language processing techniques in order to break down each word to its stem form before searching for it in the database.
Accomplishments that we're proud of
- Developed a UX Research plan and recruited and interviewed 7 users in 1 day.
- It was my first hackathon, so I'm proud of submitting a completed product.
What we learned
- Gained exposure to useful frontend libraries like Tesseract.
- Practiced using different features of Figma.
- Working with new apis like Wikipedia and Natural Language Toolkit.
- How to use data preprocessing techniques to better our algorithm.
What's next for Oneste: A Cosmetics Safety App
- Partner with health watchdogs to integrate more existing systems and create a better UX
- Add functionality to quickly access reviews for scanned products
- Finesse the backend to provide only the most relevant information from the Wikipedia API
Log in or sign up for Devpost to join the conversation.