SmartCrowd is a people tracking application designed for populated environments, leveraging advanced computer vision techniques for live analytics and status tracking, developed throughout my Computer Vision course at CMU. My role in this project was to spearhead and optimize the integration of the YOLO model for people tracking with area segmentation for status tracking, designing the back-end architecture of the app. This led to a marked improvement in surveillance accuracy and capabilities in high-traffic areas, such as gyms & coffee shops.
With wait times worsening in recent years, people's growing impatience is leading to increased frustration in unexpectedly crowded settings. Coupled with busier lifestyles, this intensifies feelings of annoyance and boredom.
SmartCrowd begins by allowing the owner to map out their space into various areas, such as sitting, ordering or waiting for their coffee.
This map is turned into a mask which when coupled with the YOLO algorithm, can output the status of each person in view of the camera, as well as live occupancy of each area.
Ever wanted a Capri-Sun but couldn't find one with flavours you like? This project makes custom pre-packaged drinks by pouring drinks into pouches and then heat-sealing them.
Sebastian Levy - Portfolio