WAWLA (What Are We Looking At)

WAWLA (What Are We Looking At) is a workshop that dives into the world of computer vision, image training datasets and the invisible human labour that makes machines see. We will get familiar with the workflow of creating a computer vision model and the steps behind assembling and preparing an image training dataset while specifically focusing on image segmentation. Segmentation is crucial for computer vision as it “points out” the parts of an image that are of value to the system. At this stage of image processing, an image is visually separated into areas of information. By outlining an object of interest (a road sign, a deer, a face) or filling in a region of particular focus (the road, the tree, shadows), these marked segments ultimately inform the computer vision model’s way of perceiving the world. What is and isn’t segmented depends on various factors, from the technology’s future application to cultural context, image resolution, operation language, subjective perception, and might differ from our own way of reading an image.

During the workshop, participants will get introduced to different examples of image training datasets and the workflow of processing visual data through image segmentation tasks while being faced with the ethical, philosophical and political challenges of teaching systematic vision and deciding “what is important to see”. This workshop is deliberately non-technical. We will be working with paper and markers for accessibility and to make tangible the labour and compromises involved in creating technology that otherwise is obscured by innovation.

Duration: 3h
No prior knowledge required.