Authors

Inspiration

The world health organization says the Coronavirus outbreak is now officially a pandemic. According to the CDC, the best prevention against the coronavirus is washing your hands thoroughly with soap and water. When we watched the hand-washing video, we realized that we missed many parts on our hands such as the backs of hands and fingertips if we didn't adopt the correct hand-washing technique. The purpose of this project is to make people learn the correct ways of washing hands and stay healthy, away from viral infections such as COVID-19 and flu.

What it does

The "Hand Virus Scanner" is for forming proper hand-washing habits by scanning your hand washing. When you upload a hand-washing video to our website, our website classifies which step you are in. And it shows how long you wash hands in each step and how much percentage your hands become clean.

How we built it

We started from researching the appropriate pre-trained machine learning model to be applied to our project. We picked the Wide ResNet-50-2 and replaced the classifier part of it to train on our own images so that the model can identify the images we give it. While training the model, we worked on web development including mobile web using Flask, Javascript, HTML, and CSS so that users can upload video files and get the analyzed results.

Challenges we ran into

The most challenging part we ran into was team management and communication, especially this hackathon was held online and we couldn't work on it at the same place. Since it was crucial to manage time to complete our project within a short period of time, we separated the tasks and put the best effort to connect each other using various communication channels such as google hangouts, google drive, and git hub. From the technical perspective, we were not familiar with most of the frameworks and libraries used in this project but we could learn by trial and error.

Accomplishments that we're proud of

Throughout the developing process, our team not only learned to use Pytorch Vision to develop ML but also learned the Google Cloud, Flask. We did back-end stuffs and also did front-end stuffs. We implemented a responsive website that runs on the server built with the Flask. This offers the proper way to wash your hands and the analyzed results based on the video the user gives. This application can contribute to prevent getting diseases from viruses by keeping the most basic rule which is washing hands.  Team Working is an important part throughout the whole Hackathon so we collaborated using several communication channels like Slack and Hangouts. Based on the strong connections between team members, we could build the application successfully.

What we learned

Throughout the development of this project, we learned more than we expected. At first, we didn't know how to design and develop our idea. However, we could learn about the concepts of convolutional neural network and machine learning as well as various skills such as deploying webpage on Google Cloud, Pytorch, and Flask. Furthermore, we learned that we can make progress on work helping each other, even though we didn't have much knowledge about it.

What's next for HandVirusScanner

We would like to detect streaming data, show streaming videos with analyzed results provided by the machine learning model.  Then, we would like to train our machine learning model with more data, so that our model can have higher accuracy on the results. For now, people can access the website through mobile web browsers on the cell phone but we're planning to develop an app to provide users more user-friendly and visually pleasing service.

Try It out

Hackathons

Technologies

flask, html5, javascript, machine-learning, opencv, python, pytorch

Devpost Software Identifier

255499