1. Was your social distance detector effective at detecting potential violations?
Although the social distance detector was able to identify individual people and detector when certain individuals got close to each other, there were many problems that were in the social distance detector. With the video being at a first person perspective, I noticed that the detector could’ve falsely detected violations when a person is overlapping another but may not be less than 6 ft away from each other. Similarly, it also detected certain individuals who would be walking literally next to each other as not violating social distancing. So overall, it was not very effective with the video that I had used.
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?
I do not think this approach would be effective for estimating new infections in real time. The program will have assume an amount of people who would be infected with COVID-19 in order to infect others when they are less than 6 ft apart. If we are using this for the COVID-19 pandemic, we could use it as a good example to demonstate how quick and easy COVID-19 could spread, as a simulator to educate the people. We could also use it to follow certain individuals that were determined to have been infected to see who else they may have infected.
3. What limitations or improvements might you include in order to improve your proposed design?
In order to ensure that the detector works accurately, I think a top-down view used for the social distancing detector would ensure that perspectives are not messing up with the detector. If possible, in an ideal world, having the program able to detect the perspective and then do its calculations through there would definitely help a lot.