Forensic Architecture

Machine learning offers new, powerful, and indeed groundbreaking opportunities to monitor and modulate socially and politically critical infrastructure. In many areas, those opportunities are already being taken, often without due consideration or due exposure, and it is as such urgently necessary to think seriously about ethical applications of machine learning in both theory and practice. It is exactly due to its increasingly extensive use in apparatuses of state surveillance and autocratic control that we practice the ethical deployment of machine learning systems; to invert that gaze and reason constructively about the technology’s place in civil society.

I.102  German Colonial Genocide in Namibia: The Hornkranz Massacre

Date of Incident
12 April 1893
Location
Hornkranz, Khomas Region, Namibia
Forums
Exhibition, Human Rights Report, Media
In Partnership With
Nama Traditional Leaders Association (NTLA), Ovaherero Traditional Authority (OTA), Forensis
Summary
On 12 April 1893, German troops attacked the Nama settlement of ||Nâ‡gâs, also known as Hornkranz, located in present-day Namibia. This was the little-known first chapter in a genocidal campaign against the Nama and Ovaherero that culminated between 1904 and 1908. We worked with descendants to reconstruct the lost settlement and support Nama claims for access to and preservation of the site.

I.7  The Forest Under Detection

Date
1492 – Ongoing
Summary
Using spatial, image, and data analysis, we investigate the history of the Colombian states' cartographic domination of the Nukak people in the department of Guaviare.

I.56  Cloud Studies

Date
2008 - Ongoing
Summary
Toxic clouds colonise the air we breathe.

I.54  Model Zoo

Date of Incident
Various
Location
Various
Forums
Exhibition, Media
In Partnership With
Bellingcat, Amnesty International, Open source contributors
Summary
Since 2018, Forensic Architecture has been working with ‘synthetic images’—photorealistic digital renderings of 3D models—to train machine learning classifiers. Model Zoo includes a growing collection of 3D models of munitions and weapons, as well as the different classifiers trained to identify them making a catalogue of some of the most horrific weapons used in conflict today.

Detecting Tear Gas: Vision and Sound

Date
20 Feb 2020
Summary
This study documents several case studies undertaken by Forensic Architecture researchers and OSS contributors. Following on from research we released in early 2019, in the course of the year we have expanded our research in human rights and machine learning, with a particular focus on applications of synthetic data.

I.46  The Battle of Ilovaisk

Date of Incident
24 Aug 2014 – 03 Sep 2014
Location
Ilovaisk, Ukraine
Forums
Legal Process, Media
Commissioned By
European Human Rights Advocacy Centre (EHRAC)
Summary
The victory of pro-Russian separatists over the Ukrainian army in the battle for control of Ilovaisk was a defining moment in the 2014 conflict in eastern Ukraine. We used machine learning to search for evidence of Russian military involvement.

I.43  Triple-Chaser

Date of Incident
26 Nov 2018
Location
Tijuana-San Diego border, US/Mexico; Gaza, Palestine; and around the world
Forums
Exhibition, Media
In Partnership With
Praxis Films
Summary
When US border agents fired tear gas grenades at civilians, including children, photographs linked the grenades to the Safariland Group, which is owned by Warren B. Kanders, the vice chair of the board of trustees of the Whitney Museum of American Art. In response to our invitation to the Whitney Biennial 2019, we developed machine learning classifiers to search for Safariland munitions online.

I.3  Computer Vision in Triple Chaser

Date
30 Nov 2018
Summary
For our contribution to the 2019 Whitney Biennial at New York’s Whitney Museum of American Art, we developed a machine learning and computer vision workflow to identify tear gas grenades in digital images. We focused on a specific brand of tear gas grenade: the Triple-Chaser CS grenade in the catalogue of Defense Technology, which is a leading manufacturer of ‘less-lethal’ munitions.

I.2  Experiments in Synthetic Data

Date
06 Nov 2018
Summary
In many investigative contexts, the abundance of potentially relevant data presents a significant challenge to constructing a case. To this end, we use machine learning tools where appropriate to channel investigative resources effectively when processing footage from YouTube, Vimeo, or other media sources.