I found the article to be very interesting, before reading I had a very primitive idea as to how a computer can recognize objects in a video or anything else to do with a computer processing a video. But, as I was reading I was reminded of examples of “computer vision” as the author put it. On the way to the Intro to IM class there is a screen and camera, and displayed on the screen is whatever is in front of the camera but the brightness of (nearly) every pixel is represented as a character. The end result is a black and white image but with characters to represent the brightness of most of the pixels. I was also reminded of a time where during the thunderstorm which had happened recently I met someone as I was walking to D2 who was having trouble photographing a lightning strike. In my head I thought that he could have benefited from some kind of program which could do all the work for him. I thought about it, and the idea I came up with was if the camera was recording all the time but deleting footage older than 30 seconds (to save space) then if the camera detects a spike in brightness it saves the last 30 seconds plus some footage after the lightning strike (this could be done by delaying the time between detecting the spike and saving the footage). Of course I don’t really know how to implement it but in theory it could work… I think.
I also learned a lot from the article. There were techniques mentioned, most of which which I would never be able to come up with myself. “Frame differencing,” “background subtraction,” and “brightness thresholding.” While I do not have a great idea of how to implement these techniques I think the most valuable thing I took away from the article were the names as I could always search them up and learn more afterwards. Fortunately they also linked some processing programs at the bottom that I could use to learn more.
Lastly, I noticed that the article was somewhat outdated. It was released in 2006. I felt it was worth mentioning because at the time machine learning was no where near as advanced as it is today so I would have liked to have learned more about how machine learning could be used to improve “computer vision.”