I found this article “Computer Vision” to be quite fascinating, especially when the author talked about how computer vision technologies used to be limited to military and law-enforcement purposes, when it is now also used for programming and creative purposes. Computer vision holds a significant position in creating ideas and designs for areas such as novel artworks, games, and home automation systems. I really liked how this article showed a lot of examples of whatever the author was talking about, such as the first interactive artwork to incorporate computer vision and elementary computer vision techniques, while also showing visual examples of the concept. This really helped me understand the idea and see how it would play out in real life. I also really enjoyed how much I learned in quite a little amount of content compared to our other readings, which I’ll talk about later.
Some of the ways in which computer vision differs from human vision as stated in the article is “generalization,” where computer vision algorithms are not fully general as they are highly dependent on assumptions about the real-world video scene. On the other hand, humans are easily able to adapt and generalize what they’re seeing across a bigger range of contexts. Another difference is computers’ vs humans’ abilities to differentiate and infer what they’re seeing if the subjects are too similar in color or brightness. Computer vision struggles to work well if there isn’t a significant contrast between people and the environments, meanwhile humans are able to tell people and things apart, even if the brightness and color of the backgrounds are very similar. These made me think of how easy it is for us to determine what things are. We’re very lucky to be able to process these things so quickly, because not only would it harm us in terms of survival, it would also harm us in terms of social life. Imagine your friend sees you and waves at you, only for you to not be able to process or see them because they’re unintentionally matching with the wall. It would be embarrassing for them and complicated for you. It also helps that we can generalize enough to be able to tell that our friends are our friends instead of processing them as a newly discovered bug just because they decided to wear a new bright color palette of clothes one day.
One more difference that is also a technique humans can use to help the computer see is by using infrared. Infrared is invisible to the human eye, but helps improve the signal-to-noise ratio of video that’s capture in low-light conditions. This helps computer vision operate in near to complete darkness, something that humans can’t quite do. Another technique is the use of retroreflective marking materials, which helps ensure high-contrast video of tracked objects due to the light illuminating and reflecting back. Reading on this unintentionally reminded me of and answered a question I forgot I had back when I was a kid in elementary school. When crossing the street to get to school, I always wondered why the crossing guards wore bright yellow or orange jackets with silver stripes. This answered why. The high-contrast brightness and reflectiveness is to enhance their visibility, especially in dark conditions, by illuminating light back to the drivers.
Before reading this article, I would think that when it comes to interactive art, computer vision’s ability to track would allow for direct engagement with the artist and/or the audience. It enables immediate user feedback from viewers’ actions, such as Myron Krueger’s “Videoplace,” which analyzed participants’ silhouettes and movements to create graphical responses in real-time based on their interactions. These real-time interactions and visual responses are able to enhance the experience and make it more immersive and engaging. I thought that it would just be for good interactive fun. But after reading this, I realized how many issues could rise from it. It could end up being problematic, like Bureau of Inverse Technology’s “Suicide Box,” which has led to controversy over the ethics of recording suicides and if the suicides are even real. Even with the bad and good, it could also be used in a negative light to help shed light and raise awareness. David Rokeby’s “Sorting Daemon,” inspired by concerns over surveillance and profiling, uses computer vision to survey the environment and extract and sort people in his piece as a way to question the ethics of using technology for reasons such as profiling. I didn’t realize how complex the uses of computer vision could get, but now I’ve seen several perspectives on how it could be viewed as fun, beneficial, or problematic.
Overall, this was a really good read and I learned a lot more than I expected from it. I thought I would just learn what computer vision is and a couple of examples of how it works and how it could be implemented into interactive art, but I ended up learning way more than that