Mini Project 1
02/14/2021
Videos
Question 1: Was your social distance detector effective at detecting potential violations? Are you able to describe how the distance detector is applying its calculations of either being safe or noting a violation?
The social distance detector was fairly effective at detecting potential violations. When people in the video started walking closer together the rectangle around them turned from red to green to signify that they were too close together. I don’t understand all the details of how the distance detector is making its calculations and coming to its decision, but I think I have a general idea of what some different blocks of code are doing. At the top of the code, different modules are imported that are utilized later in the code. Then different functions are defined that perform many different tasks. For example, I think the Check function determines if 2 people are far enough apart and returns True if they are and False if they aren’t. The ImageProcess function is called later in the program and obviously does some processing, but I don’t understand exactly how it works. I believe the code essenitally goes through the entire video frame by frame and determines where each person is located and then whrther or not they are too close to any other identified person. Then, if they are too close the box is drawn in red around them and if they are not too close, the box is drawn in green. These boxes are placed over the original imags and then put together to form the output video that is created and placed in the output-videos folder.
Question 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 feel like the general approach could be effective for real-life infections. The strategy of determining who was in close contact with who is central to figuring out exposure levels of different individuals. If we were able to put up cameras in public places, especialy in highly populated areas, we could collect a lot of video data that could be used as inputs into a model like this one and could be effective as an approach to tracking infections in response to the COVID-19 pandemic. It would also be important to try to make the code as efficient as possible so that it takes less time to process since we would have to sift through a lot of video footage.
Questian 3: What limitations or improvements might you include in order to improve your proposed design?
One problem I forsee is difficulty acquiring an adequate amount of video data. I also am not sure if there would be ocnfidentiality concerns when taking video in public places. Additionally, there are other factors that impact infection rates beside physical distance apart such as whether or not individuals are wearing masks, so it would be better if the model could take that into consideration as well. Also, as I mentioned earlier, it would be nice is the model could sift through the video more quickly because it took my model over 5 minutes to get through a 30 second video.