Shel.io

Our app takes in a video stream, which can be a previously recorded video or webcam live stream, and then uses the Mask R-CNN model within the Detectron2 library to perform a semantic segmentation, pixel-by-pixel labeling of an image, to generate larger labels of areas (people). From there, we count the live number of people in a given area and determine whether or not the current density should be addressed before the situation becomes dangerous. These density calculations take into account the room type and the room’s square footage, both of which can be input by the user. The application then sends a mobile notification to a user if the occupancy level passes a certain user-specified threshold using a notification app called NotiBot.

I created this project as part of a team at Bitcamp, the University of Maryland’s premier hackathon. I find much fulfillment in solving social problems with code and using technology for social good. This was my inspiration when I approached a group of fire marshalls working at the Bitcamp event and asked them about the challenges they face at work. The 20-minute conversation that followed is what led to this project.