The reading changed how I think about the difference between computer vision and human vision. Humans can recognize a person or object even when the lighting, angle, or background changes, because we understand context. A computer sees pixels and must follow programmed rules, so it can easily become confused. I think I can say that I have experienced this when my phone’s face recognition failed when in poor lighting even though another person could recognize me immediately. We can help computers track what interests us by using stable lighting, a plain background, strong contrast, colored markers, brightness thresholding, frame differencing, or background subtraction. I agree with the author that improving the physical environment can sometimes be easier than writing more complicated code. This made me realize that designing a computer-vision project involves arranging the camera, lighting, and space as well as the programming itself.
I also think the author is slightly biased toward the positive and creative possibilities of computer vision because he is an artist and practitioner in this field. Although surveillance concerns are mentioned, I feel the reading could discuss privacy and consent more strongly. Tracking can make interactive art exciting because people can use their bodies instead of a mouse or keyboard, but the same technology can record people who did not agree to participate. This could make an artwork feel uncomfortable instead of interactive. If I used computer vision, I would clearly tell participants what the camera tracks and avoid saving their video.