Reading through the passage by Golan Levin and others, I was re-introduced into the world of computer vision without being overwhelmed by technical jargon. As someone who used to get in touch with CV only on a library-resorting basic, aka. using OpenCV in Python, rediscovering some rudimentary algorithms behind it from a perspective of technology in arts brought me some really fascinating ideas compared to the old-school path-finding tasks.
The paper acknowledges how CV has traditionally been the realm of expert researchers and applications limited to military and law enforcement (and yes, of course, teenagers’ robotics tournaments). With advancements in software development tools, the rise of open-source communities, and the affordability of digital video hardware, CV is no longer an exclusive playground. Artists and designers are now empowered to experiment and innovate, integrating vision-based interactions into their creative projects with relative ease.
A particularly insightful section for me is to emphasize the importance of the physical environment in enhancing the performance of CV systems. The authors argue that optimizing lighting conditions, using retroreflective materials, and selecting appropriate camera lenses can significantly improve the accuracy and reliability of vision algorithms. In my own projects that utilized webcam and process on the video pixels, although the ‘algorithm’ is simply resorting to the grayscale conversion, it is already noticeable to see how the physical environment affects the outcome. The collaboration between software and the physical setup underscores a holistic approach to designing interactive media, where both technological and artistic considerations play crucial roles.
Reflecting on the questions posed by the paper, it becomes clear how CV fundamentally differs from human vision. While humans effortlessly interpret context, recognize patterns, and infuse semantic meaning into visual data, CV relies on meticulously crafted algorithms to process pixel information without inherent understanding, creating both opportunities and challenges for us. By all means, the opportunities lie in the possibility of fine-tuning, altering, and utilizing what the computer has ‘seen’ with algorithms (in other words, to ‘think’ after the ‘listening’ and before ‘speaking’) while it is hard to change the optical structure of human vision. On the other hand, this ‘thought-precedes-realization’ approach in harnessing CV could result in hindering the imagination of artists as the fine line between envisioning and then realizing and the more intuitive manner of mixing the creation and designing at the same time can be easily blurred by lines of codes.
Besides, this powerful capability also introduces ethical considerations, especially when used in interactive art. The capacity for tracking and surveillance can enhance the immersive quality of art installations, making them more responsive and engaging. Yet, it also raises concerns about privacy and consent. For web-based instances like the games I’m currently working with, it is easy and necessary to ask for permission from the user, while installations and broader observational systems could skip the users’ first-hand consent. How do we balance the creative potential of CV with the need to respect individual autonomy and privacy? These questions are crucial as artists and technologists continue to push the boundaries of what interactive art can achieve.