What are some of the ways that computer vision differs from human vision?
Computer is really context dependent compared to humans. We have eyes and can generally differentiate objects and perform any actions with any inputs, but no computer vision algorithm is completely autonomous. Each algorithm is dependent on its code and assumptions about the specific scene it is analyzing. If conditions such as absence of movement or poor lightning are present, then algorithm may fail.
What are some techniques we can use to help the computer see / track what we’re interested in?
As mentioned in the abstract, we need to increase the contrast more so the computer vision can differentiate between environment background and people’s movements. Those include lighting that silhouettes people, contrasting costumes. Also, using Infrared Illumination improves signal-to-noise ration in low-light conditions and retroflection marking materials.
Choosing the right imaging hardware is essential too. For example, telecentric lenses so object’s magnification is independent from the distance, polarizing filters to reduce glare from reflective surfaces, and very purposeful choice of cameras for high resolution, frame rate, short exposure, dim light, UV light, or thermals.
How do you think computer vision’s capacity for tracking and surveillance affects its use in interactive art?
It is, for sure, a core mechanism and an engine that gives power to interactive art. I strongly believe that computer vision’s ability to detect, track, measure presence, motion, color, and size completely changed how interactive art is perceived since it’s invention. Techniques such as background subtraction of frame differencing, even though simple, are very profound in how they helped to make thousands of modern interactive art installations. For example, advanced tools like EyesWeb specifically focus on the tracking and surveillance and provide ‘analysis and processing of expressive gesture’. Now, it is not just about detecting movement, but rather interpreting for specific musical or visual artistic purposes. I also think that first interactive piece called Videoplace that I read about on my other IM class gives audience agency and computer vision acts as a bridge between human input and computer’s further output much like a computer mouse, but rather detects human movement and gestures.