1. Automated Contact Tracing using Person Tracking and Re-Identification
- Author
-
Aditya Kotwal, Vedant Wag, Abhay Khandelwal, and Pratham Sutone
- Subjects
Coronavirus disease 2019 (COVID-19) ,Computer science ,business.industry ,Process (computing) ,Confusion matrix ,Computer vision ,Artificial intelligence ,Tracking (particle physics) ,Precision and recall ,business ,Protocol (object-oriented programming) ,Re identification ,Contact tracing - Abstract
This paper presents an implementation of contact tracing (a protocol followed to curtail the spread of viruses like Covid-19) using CCTV video footage from multiple cameras. The proposed system gives immediate insights about all the possible people at risk when a person under surveillance is clinically tested positive. This system automates the process of manual contact tracing by detecting contact between people under surveillance by estimating the distance between them, identifying them, and tracking their interactions. This data is then stored, filtered, and analyzed. A confusion matrix was derived from the videos that took into account true contacts, false contacts between people in the input video and whether the contacts in the video were considered as true contacts or false contacts according to the algorithm. This gave us insights about the algorithm’s accuracy, precision and recall which were found to be 91%, 94% and 71% respectively.
- Published
- 2021
- Full Text
- View/download PDF