One way computer vision differs from human vision is in which computer vision uses sensors, which is almost similar to the human eyes, except their ‘vision’ is not an image but more like bits of information. Though both ‘visions’ need to be understood, the process for a computer would be longer and more complicated in which information has to be internalized and run through a series of program.
To help the computer see we must make the object of attention to be easily visible by either having it be a very bright light, so the computer can track the brightest pixel, or have it in an environment that contrasts the desired object to be tracked/seen (like against a green screen)
As much as it is interesting and cool, the idea of computer vision in relation to surveillance will always come with privacy concerns, especially as seen with the Suicide Box project by the Bureau of Inverse Technology. I think there should be some type of consent from both the artist and those that interact with the art. However with art like the Suicide Box, there is no consent and it all just seems like an invasion of privacy.