As far as the global pandemic is concerned, the novel coronavirus, aka COVID-19, has brought the world down to its knees. Regardless of the rich or poor, the virus continues to put an adverse impact; not only taking a severe toll on the physical and mental health of individuals but the economic health as well. However, it is the poor that is being affected the most. Talking about my homeland, Pakistan, people have met with the same fate as the people who live across the globe. This is what inspired me and my team to come up with a solution that helps curb the adversaries of the virus and hence proves beneficial to society.
What it does
Wanderlust is a set of programs designed to help analyze and track individuals coming in contact with a COVID-19 patient. This is performed by looking at their location history to provide them with alerts for the maintenance of social distancing. The solution is devised keeping in mind the educationally and digitally deprived population. Wanderlust uses WhatsApp-based and a camera/CCTV-based infrastructure to track the location and images of identifiable individuals. The system predicts with near accuracy the individuals who came in close contact with the affected and alerts them to impose self-quarantine before it's late, hence help flattening the curve.
How I built it
The tech stack comprises the following:
Python, C, MySQL, Microsoft Excel, Linux, Ubuntu, GNU Bash, NumPy, OpenCV, scikit-learn, Twilio, WhatsApp, etc.
Machine Learning (ML) and Deep Learning (DL) models have been deployed to help create a solution.
Challenges I ran into
One of the challenges I and my team ran into was the selection of a service that delivers programmable messages. We decided among Twilio, Plivo, and Infobip, and picked Twilio for our tech stack due to its agility.
Another challenge was that the desired prediction accuracy was not being achieved. However, this was solved by my teammate Rafey, as he appropriately split the train and test data to meet the desired output.
Accomplishments that I'm proud of
I and my team feel immense pride in devising a solution that fulfills our goals and objectives. Current solutions are half-cooked and limited to the computation of the distance maintained between two individuals with no appropriate alert mechanism that ensures prompt action. On the other hand, WanderlustAI provides a complete solution ranging from computing the distance to the analysis of data and SMS and WhatsApp notifications.
What I learned
Before incepting into this project, I didn't know that a simple tweak of splitting train and test data would solve the inefficient output issue. Thanks to Rafey for helping me out.
What's next for WanderlustAI
The project is currently a prototypical version. Efforts are being made to escalate from the prototypical phase to the finished product phase.