songmin17 / vacant_seats_detector

Vacant Seats Detector

"Vacant Seats Detector" is a computer vision project that finds empty seats at crowded places, such as restaurants and cafes. It creates a layout of the room in bird’s eye view and produces a real-time visual depiction of table capacity and table occupancy. Through a pipeline that consists of object detection via YOLOv7, conversion of image features to and from world coordinates, and feature detection through k-means clustering, we produce an accurate graphical output of the place. By placing a video of this depiction at the entrance of a diner or on a website for the diner, customers can easily check for open seats.

Report

For more information regarding the details of the project, please refer to the report in the repository.