There are many daily habits that are detrimental to the environment, such as the use of plastics, the use of disposable products, and the waste of materials and energy. Our goal is to create a solution to recognize those habits and provide a better alternative.
What it does
An object detection web App that recognizes environmentally unfriendly habits and provides feedbacks
How we built it
We used React to build a component-based single page web application that integrates with Google Cloud Vision API and configured with Firebase Storage and Realtime database. Our app also uses Microsoft Azure as a Web Service so that it runs within a single resource group
Challenges we ran into
React framework is an evolving framework and therefore two of our team members who had experience with React still need to adapt the changes. Two of our team members are new to the web application and had little experience with real-world applications and it took us some time to get everybody on the same page. There is some time-consuming problems while we were dealing with firebase storage and database. We also ran into problems on how to correctly convert the data URI to file or to Base64 and vice versa.
Accomplishments that we're proud of
We integrated the Google Cloud platform API to our application, specifically the Google Cloud Vision service. Our web app is running on Azure web services and stores each snapshot into both firebase storage and real-time database.
What we learned
Training a problem-specific machine learning model(versus a general problem-solving model ) will enhance the ability our of detection program. We did learn a lot about modern machine learning and computer vision technologies and how to apply them in our project. We are also now confident with our web programming skills, especially how to build a project using React. Finally, this is our first experience with firebase and we did learn a lot from this world-class infrastructure.
What's next for Kusa
Finish the "Get feedback" feature which gives the user an overall evaluation on a daily or monthly basis is definitely what we should focus on next. User authentication and database are also definitely two features that we can add to our existing project. A more advanced and intelligent detection program will also be beneficial for our overall final evaluation. There can also be more user-friendly features that we could implement including the integration with home assistant devices.