Skip to main content

Canteen Installation

Posted: 11/06/15

Soon after I had watched Stanley Kubrick’s The Shining for the first time, I watched Room 237, a documentary about possible meanings and different interpretations of the movie. Especially Kubrick’s deliberate use of subliminal images, intentional contradictions and continuity errors, that the viewer often only perceives unconsciously, and their disorienting effects, is what I found particularly fascinating. Our heads are so busy following the storyline about the family’s fate that we do not notice a number of major contradictions: e.g. the manager’s office has a window with daylight shining through even though it is entirely surrounded by other rooms, a chair suddenly disappears in a shot, the typewriter changes its colour etc. The list goes on and on.

This made me think if someone who is less concerned about emotional relationships, but more about logic, would be a more suitable reviewer. Inspired by Matthew Plummer-Fernadez’ Novice Art Blogger I took a frame every two seconds from certain scenes and sent the stills through an image recognition software which is based on a deep learning algorithm developed at the University of Toronto.

Unsurprisingly the software faces further challenges besides a correct description. It has absolutely no clue about the meta-context of the movie or any cultural understanding, not to mention Kubrick’s work. This is partly because of the database it takes its comparisons from, but it results in quite amusing descriptions.

Furthermore it treats each frame as an individual standalone image, not as an image taken out of a sequence of images. The software does not compare between the previous and the following image, and therefore cannot draw conclusions on a potential storyline. I assume that meta-context and image sequences are very likely going to be the next big goals in the development of image / film recognition software, and due to the exponential growth of computing power might be achieved much sooner as we expect.