Everyday millions of people around the world struggle with their skin, and more importantly, struggle to understand what is going on with their skin. With DermaCare, our team aims to provide an educational and informative platform for those who struggle with any and all skin-related issues.
What it does
Specifically, DermaCare revolves around Google's AutoML Vision AI API. This customized machine learning model classifies numerous sets of images which are categorized by skin-types. Once a user uploads an image through the website, built on pure HTML, CSS and integrated with the Google Cloud database, the Machine Learning Model analyzes the image and returns a result of the type of skin, along with a report including the symptoms, causes, and a possible resolution and more. Currently, however, the Machine Learning Model is particularly tailored to skin issues related to acne, rosacea, lyme-disease, and eczema.
How we built it
The application was built through the VSCode IDE. Python, accompanied with, the Flask framework, were used to build and maintain the backend of DermaCare. The Flask framework helped in integrating the external API sourced from Google's own API's, as well as, connecting the Google Cloud Database to deploy the actual application from our machine to the web as a functional application.
Challenges we ran into
Some Difficulties during development were setting up our AutoML project and finding usable data to train on, and deploying our Flask based application with our AutoML project with python to a user-friendly web based environment.
Accomplishments that we're proud of
As a team, we are very proud of our ability to have built a customized and built an accurate (for the most part) machine learning model for our first time. We also pride ourselves in our ability to have actually completed and deployed a program into a web application environment within a constraining time period.
What we learned
Throughout this process, as a team, spec wise we have collectively learnt how to utilize and integrate external API's, optimize Google's cloud services, and how to develop a dynamic web application with HTML and CSS. However, we also have learnt what it means to truly work as a team. The importance of effective communication, organization and time-management prevailed throughout this learning process as our process of thinking, to building, and deploying would not have been the same without each of these components.
What's next for DermaCare
We plan to optimize our Machine Learning Model by expanding our database to recognize a wider range of skin needs by obtaining and training more data sets. We also plan on creating a more dynamic and aesthetically pleasing application by implementing new features such as a personalized quiz in relation to the skin issue, adding more CSS details, and providing more accurate and thorough information.