Data 310 project summaries
1. Was your social distance detector effective at detecting potential violations?
Yes, however, this was not the case with all videos. I noticed that it identified violations much better when the angle of the camera high but not directly overhead.
2. Do you think this approach would be effective for estimating new infections in real time? How would you implement such an approach in response to the COVID-19 pandemic we are currently experiencing?
No, this approach would not be effective for estimating new infections in real time. Research has shown that rate of infection in well-ventilated areas where people as moving around to be negligable. Additionally, there are other influencing factors such as an individual’s immune health that help determine if they will contract the virus. Perhapse if this technology was used in schools (or shopping centers when they reopen) were students may pass eachother in unventilated hallways or sit close to eachother in class, it could be used to track exposer of students to a known carrier (that is, if a student tested positive, we could track all the students who had come in contact with them).
3. What limitations or improvements might you include in order to improve your proposed design?
The camera angle was a limitation in many of the videos I tested the model on. In the future, the design could be improved with an input of the camera angle and height that would scale the number pixels that represent 6 feet in different areas of the frame. Additionally, videos could be captured from directly above to reduce the issue of perspective in determining peoples’ locations relative to the camera.