Denise Scott Brown and Robert Venturi, influential architects in urban planning, stated in Learning from Las Vegas that „learning from the existing landscape is, for the architect, a way of being revolutionary“. Such learning is possible through numerous tools: drawing, photography, and data collection – but perhaps one of these tools is the most effective of all, which is, as both architects suggest – the gaze.

This artistic project explored the federal state of Bremen to map the visible and invisible landscape. The first, visible, was carried out along a route of just over 45 kilometers in which the territory was photographed in its specificities; the second, on the other hand, was mapped through a reader-object (Image 3) produced in order to suggest landscapes that were left out of the machine learning circuit – which was therefore taken out of our reach.

To highlight the landscapes of the territory that are not detected by machine learning processes, an application written in Python was developed that processes two important stages of this artistic project: firstly, it highlights and removes from the image the content within the meaning boxes – selections of specific areas of the images that contain labeling. To do this, the application makes use of the Google Vision API – a service offered by Google that allows developers to integrate computer vision capabilities into their applications. The result of this process is an image in which, instead of seeing what the machine learning process sees, i.e. the content inside the meaning boxes – we see the rest; or rather, what has been left out and which is not important for computer vision (Image 1 and 2).

Image 1

Image 2

In the second stage, the application provides the user with the area left out by the machine learning models as raw material, or in other words, as ink kind of matter- to be used by the user using a stylus pen to draw and envision possible other landscapes that were taken out of reach (Image 3).

Image 3

Drawing comes into play here as a method of imagining what we could see. Drawing, as a mental phenomenon (until the mid-17th century, the concept of „drawing“ in Portuguese was synonymous with having an idea, drawing in thought), is a cognitive process: to draw things is to find a possible meaning behind what we see – in other words, to find blind spots in our perception.

The images used in the first stage of the application‘s processing are those uploaded and georeferenced on Google Maps every day by thousands of users – this is because such images today populate our imagination of the territory. The reader-object captures these images in real-time, removes the meaning boxes, and offers the user the remaining pixels for the user to glimpse other territories. Finally, through a poetic-imaginative action, this artistic project, therefore, envisages thinking about how these socio-technical processes can be reimagined to promote more radical ways of seeing the world through machines – as well as questioning how training sets can be redesigned to act on our ability to see and engage politically in the land-scape.

Installation at the Master Project 2024 exhibition. A 9-meter line of Undetected Landscape was installed in the space.

At the Master Project 2024 exhibition, visitors could draw territories using georeferen-ced images on google maps.

This publication is the result of walking a little over 40 kilometers between the two extremities of the state of Bremen (starting in Bremerhafen and ending in Mahndorf) with the reader-object in hand. The result is a long strip of landscape where you can see the other hidden territories.