Pixels without Vagueness
Pixels Without Vagueness (Version 0.9)
Digital video, Animation made with Open CV face detection (Python, written by Ashwin D'Cruz).
The video poetically explores the gap between human perception and the emulated perception of a computer vision algorithm based on Artificial Intelligence.
The narrative utilises the Sorties Paradox (also known as the paradox of the heap) as a means to point out a kind of semantic vagueness humans may see that a computer does not see (in this case a face detection algorithm). To demonstrate this gap, the video presents an animation that counts a minimum number of pixels which an algorithm need in order to detect a face in an image. The exactness and granularity of the discrete point of detectability suggest that the algorithm lacks in sensing a type of ambiguity or nuances that we may find in human perception.
The video questions whether the quantified nature of the discrete process of seeing (and not seeing) by a machine dehumanises a human subject. Today, processes of algorithmic recognition and filtering can decide whether a subject is included or excluded for our social and technical systems. Either one is excluded or included for the systems, could some kind of dehumanisation occur, as the systems may not recognise humans or the systems may overlook very human features of the people included?
Machine Learning Engineer and Adviser: Ashwin D’Cruz
Research Adviser: Chris Williams
Narrative Adviser: Amanda Dean
The narrative adapts the terms of signal and noise by Hito Steyerl.
Streamed on the channell https://www.goingaway.tv between November 2019-March 2020 as part of The Wrong – New Digital Art Biennale X arebyte Gallery. In turn the video has been exhibited physically as part of Going Away TV curated by Marc Blazel and Alexander Harding at arebyte Gallery between 5-14 March 2020.