Perhaps the ways that computer vision differs from the rest of us humans can be the fact that most of the time, a computer requires itself to be either coded within its software or be connected to a sensory tool, such as a camera, in order to see and analyze its surroundings; unlike a human who already has the ability to analyze their surroundings much more faster and quicker. In addition, the computer will probably have to be connected to a certain engine, such as either a GPU or AI-powered processing system, in order to assist itself in being able to remember a person’s movement while keeping track of their movement at the same time. In our case, the ability to analyze somebody’s movement, whether it be in front of us or not, is already ingrained in our brain to do otherwise. To put it simply into different words, it would seem as if the computers require to have their hand be held and learn the patterns of what to identify and/or be on the lookout for while the humans have a higher chance figuring out quicker what is in front of our eyes with little to no hints needed. Matter of fact, one might say that the computer demands more manual work in preparation of “environment analysis” than us.
Now, if there must be a way for a computer to see and track something based on what we are interested in for it to watch over, there might be some techniques, as inspired from the book, that can be of assistance. The first technique to come in handy is teaching the computer to know the difference black and white, depending on which of the two is the background and which one is the object passing by said-background. This is because of how I believe that in order to help the computer be familiar with the colors of its surroundings and be able to tell the difference between certain color changes, it is often best to have the computer start from small beginnings and train it to practice telling the difference black and white; for example, if the background were to be black and the computer understands that, then it surely won’t be hard for the computer to detect a passing-by object that is the opposite of black, which can somebody in light-colored clothing that the computer may be able to tell and keep track of their movements. Another technique that may be followed by this mentioned one is the ability to identify differences between certain color shifts whenever something gets added to their image. What that means is the computer being able to remember what an original background looks like by default, only to then become capable of identifying when something enters the original image that is indicated by a color shift/addition as evidenced through anything changing within the pixels that the computer is supposed to take note of. For instance, let’s just that it takes the memory of an image featuring an empty office room, only for a security guard to enter said-office; this is become the computer’s obligation to look at both the original empty image and the new one with the guard, and then conclude that there’s been a change in color feature/pixels within the new image and now it must turn its attention onto the new person that had entered. This is all simply about modifying the computer’s vision into being to possess the knowledge of certain colors and how to pay attention to any changes that would come through the form of alternative color and/or pixel switch-ups.
With all of this being said, I personally would like to imagine the computer’s vision affecting its usage in the interactive arts with its tracking/surveillance capacities due to the fact that any works that involve the vision are established to be something unique since the works have the observers become the artist behind this type of complex digital artwork. After all, all the observers have to do is just stand somewhere in between a screen and spot where a camera/sensory tool connected to the computer can see them, and the screen will be able to not only reflect the silhouette of somebody, but also be able to conduct fluent creations of artistic patterns based on the movement of said-person. Now, of course, the capacities are not to be underestimated, for the computer’s vision requires absolutely specific definitions of lighting, background, and how to respond to and follow any changes that are most important to catch onto. Without specific definition of their target surroundings and routine of responding to probable change, the best the computer vision will do is watch somebody move their body but nothing interesting happens other than a shadow being displayed on a screen; a moment like that simply calls for some form of uniqueness in order to add something of value to the artwork plan as it must be executed by the computer vision. Overall, this is just all of my thoughts that I’ve gathered from the reading.