Image/Method: Alexey Yurenev
[Extracting gesture through machine vision]

1. Could you briefly describe the conceptual process you follow in your work?

In Manual Labor, the process did not begin with a defined concept. I started from a technical and aesthetic curiosity: how machine vision would detect and translate gestures. I used ChatGPT to generate Python scripts and began testing gesture estimation on recorded hands at work.

Only after reviewing the outputs did the conceptual layer emerge. By reading the overlays and isolating gesture traces, the project shifted from experiment to inquiry—allowing the results to shape the questions rather than the other way around.

2. Is there a dynamic movement/shape/pattern or even rhythm that you are more drawn to in your visual thinking while working?

What interested me was the reduction of the body to detected motion. When the background was removed, gestures became detached from context—almost like residual marks.

There is a movement from presence to abstraction: the worker disappears, and only the machine-recognised trace remains. I was drawn to this transformation—to the rhythm of repeated gestures translated into synthetic lines.

3. Has the image(s) you have chosen here to share shifted your understanding or tapped into an area that was unknown to you? And if so, could you loosely link an idea with an imagistic aspect?

Yes. I realised that the detected hand is not an individual hand but an aggregation—a synthetic construction trained on millions of examples. The image revealed that machine vision does not simply “see” labor; it reconstructs it through learned patterns.

This shifted my understanding of automation. The gesture trace became a way to think about the hand as a form of automation itself—and about tools as further extensions of that automation.