Caltech Studies (paintings)
Caltech 101 was the first computer vision dataset containing hand-drawn object outlines. A milestone in the development of artificial intelligence systems, teaching machines about the world we live in. A beautiful gesture of a small team of scientists around Fei-Fei Li who later went on to establish ImageNet.
Hawksbill 50 & Butterfly 3, PAIN && PLEASURE, DAM Projects Berlin, 2025
_Large.jpg)
Wild Cat 19, The White Show, Nguyen Wahed Gallery, NYC, 2025
Pigeon, 60 x 60cm, Pigment paint on Aludibond, 2024
Butterfly 3, 60 x 60cm, Pigment paint on Aludibond, 2024
Wild Cat 19, 60 x 60cm, Pigment paint on Aludibond, 2024
Flamingo Head 30, 60 x 60cm, Pigment paint on Aludibond, 2025
Crocodile Head 3, 60 x 60cm, Pigment paint on Aludibond, 2025
Pigeon 1, 60 x 60cm, Pigment paint on Aludibond, 2024
Dalmatian 13, 60 x 60cm, Pigment paint on Aludibond, 2024
Ibis 11, 60 x 60cm, Pigment paint on Aludibond, 2024
Dolphin 50, 60 x 60cm, Pigment paint on Aludibond, 2024
Hawksbill 50, 60 x 60cm, Pigment paint on Aludibond, 2024
Exhibition view, Caltech Studies (right), PAIN && PLEASURE, DAM Projects Berlin, 2025
Caltech Studies represents a critical juncture in the evolution of machine vision. It marks
the
first systematic attempt to teach computers to "see" through a process where three scientists meticulously
traced the contours of objects in over 9,000 images. This manual annotation – the literal act of pointing
and declaring "here it is" – constitutes a fundamental moment of pedagogy between human and machine,
establishing the foundation for contemporary object recognition and artificial intelligence systems.
The particular significance of this historical dataset for my artistic practice lies in its methodological
approach of manual tracing using a computer mouse. For the past decade, my work has centered on an intensive
exploration of mouse movements as an artistic medium, examining them through the lens of information
aesthetics. The decision to now work with others' lines – specifically those drawn within a scientific
context – rather than my own, marks a significant evolution in my creative process: a movement from interior
to exterior, from personal to collective gesture.
A critical dimension of my investigation emerges in the inherent imprecision of these hand-drawn contours.
These "errors" transcend mere technical inadequacies to manifest as bias – an imprint of the dataset
creators' identity and perspective. This original bias propagates throughout the development of machine
learning, multiplying across derivative datasets and becoming an inextricable component of the system.
This observation leads to a fundamental critique of digitization and machine learning: in their statistical
essence, these systems aspire to a kind of "data democracy," a averaging and summarization that inevitably
eliminates peculiarities and anomalies. As a practitioner who has long operated at the intersection of
digital technology and artistic expression, I perceive this as a crucial loss: the reduction of vibrant,
unpredictable reality to digitally approximated models.
My project thus positions itself as a multidimensional reflection: simultaneously an homage to pioneering
work in computer vision, an examination of human gesture's role in artificial intelligence development, and
a critical interrogation of digital approximation's limitations. It seeks to render tangible the tension
between technological progress and the irreplaceable value of "authentic experience" through artistic means.
The work situates itself within the discourse of post-digital aesthetics while engaging with questions of
algorithmic bias and the materiality of digital processes. Through this investigation, I aim to problematize
the relationship between human annotation and machine learning, revealing the subjective traces within
seemingly objective systems.
Caltech Studies (realtime-eink)
Caltech Studies (realtime e-ink), PAIN && PLEASURE, DAM Projects Berlin, 2025
Marcel Schwittlick’s “Caltech Studies (realtime e-ink)” (2025) is one implementation of his
work on Caltech 101, the pioneering computer vision dataset (est. 2003). This collection of 9000 images has
unique metadata assigned to them, in particular hand-drawn object outlines.
The artwork manifests as layers of outlines, each extracted from a dataset containing hundreds of images of
a targeted subject. These iterative strokes reflect an evolving process of learning, unlearning, and
relearning—an endeavor that echoes the very nature of human cognition. The machine’s repetition of lines,
its reconfiguration of forms, and its gradual refinement of visual understanding evoke a poetic struggle
reminiscent of human trial and error. The work does not simply depict an object; rather, it reveals an
ongoing process of perception and reinterpretation, as if the machine is engaged in a perpetual act of
searching for meaning.
At the heart of “Caltech Studies (realtime e-ink)” is Schwittlick’s deliberate imposition of an external
context onto his creative process. He constructs an environment where technological constraints define
artistic expression, using a pixelated cursor as a brush, an e-ink panel as the canvas, and generative
software as the machine’s cognitive core. Within these parameters, every mark made is governed by an
algorithmic intelligence, yet the constraints set by Schwittlick infuse the machine’s process with a
human-like sensibility. By restricting the machine’s decision-making within predetermined boundaries, he
introduces a dialogue between human intentionality and algorithmic execution, effectively humanizing what
would otherwise be an impersonal digital operation.
Running on a 43x73cm e-ink panel driven by custom software. The e-paper material is
similar to a classic computer screen, nearing 4k resolution but it doesn‘t emit light, is very energy
efficient and the image remains even when unplugged.
Exhibition view, Caltech Studies (realtime e-ink), PAIN && PLEASURE, DAM Projects Berlin, 2025
Caltech Studies (Plotter drawings)
All these plotter drawings have been created during the four days of the Chaos Communication Congress 38c3 in Hamburg, Germany end of 2024. Created using vintage plotters and original plotter pens from the 80s.
Caltech Studies on 2x HP7550 and 2x HP7475 plotters, created during the 38c3 (Chaos Communication Congress), 2024
Exhibition view, Caltech Studies plotter drawings (right), PAIN && PLEASURE, DAM Projects Berlin, 2025
Caltech Studies plotter drawings (38c3), PAIN && PLEASURE, DAM Projects Berlin, 2025
Cougar Body 3, 42 × 29,7 cm, plotter drawing, 2024
Gerenuk 10, 29,7 × 42 cm, plotter drawing, 2024
Rooster 40, 29,7 × 42 cm, plotter drawing, 2024
Gerenuk 25, 29,7 × 42 cm, plotter drawing, 2024
Bass 1, 29,7 × 42 cm, plotter drawing, 2024
Crayfish 29, 29,7 × 42 cm, plotter drawing, 2024
Crocodile 1, 29,7 × 42 cm, plotter drawing, 2024
Butterfly 3, 29,7 × 42 cm, plotter drawing, 2024
Butterfly 16, 29,7 × 42 cm, plotter drawing, 2024
Butterfly 26, 29,7 × 42 cm, plotter drawing, 2024
Butterfly 66, 29,7 × 42 cm, plotter drawing, 2024
Hawksbill 8, 29,7 × 42 cm, plotter drawing, 2024
Hawksbill 28, 29,7 × 42 cm, plotter drawing, 2024
Hawksbill 50, 29,7 × 42 cm, plotter drawing, 2024
Hawksbill 81, 29,7 × 42 cm, plotter drawing, 2024
Pigeon 1, 29,7 × 42 cm, plotter drawing, 2024
Pigeon 7, 29,7 × 42 cm, plotter drawing, 2024
Pigeon 20, 29,7 × 42 cm, plotter drawing, 2024
Pigeon 31, 29,7 × 42 cm, plotter drawing, 2024
Buddha 8, 29,7 × 42 cm, plotter drawing, 2024
Rooster 15, 29,7 × 42 cm, plotter drawing, 2024
Rooster 36, 29,7 × 42 cm, plotter drawing, 2024
Rooster 45, 29,7 × 42 cm, plotter drawing, 2024
Platypus 9, 29,7 × 42 cm, plotter drawing, 2024
Platypus 13, 29,7 × 42 cm, plotter drawing, 2024
Platypus 20, 29,7 × 42 cm, plotter drawing, 2024
Gerenuk 22, 29,7 × 42 cm, plotter drawing, 2024
Gerenuk 1, 29,7 × 42 cm, plotter drawing, 2024
Ibis 18, 29,7 × 42 cm, plotter drawing, 2024
Panda 16, 29,7 × 42 cm, plotter drawing, 2024
Wild Cat 33, 29,7 × 42 cm, plotter drawing, 2024
Water Lilly 1, 29,7 × 42 cm, plotter drawing, 2024
Llama 18, 29,7 × 42 cm, plotter drawing, 2024
Leopards 26, 29,7 × 42 cm, plotter drawing, 2024
Laptop 1, 29,7 × 42 cm, plotter drawing, 2024
Caltech Studies (braille)
Exhibition view, Caltech Studies (braille), The White Show, Nguyen Wahed Gallery, NYC, 2025
Wild Cat 5
braille embossing
29.7x42 cm/11.69x16.53in
Crocodile 1
braille embossing
29.7x42 cm/11.69x16.53in
Elephant 19
braille embossing
29.7x42 cm/11.69x16.53in
Dolphin 47
braille embossing
29.7x42 cm/11.69x16.53in
Ant 35
braille embossing
29.7x42 cm/11.69x16.53in
Rooster 44
braille embossing
29.7x42 cm/11.69x16.53in
Exhibition view, Caltech Studies (braille), The White Show, Nguyen Wahed Gallery, NYC, 2025
Using traditional braille printing technology, Schwittlick transforms the scientists' mouse
movements into tactile compositions through the mechanical precision of the braille embosser - typically
used for transcribing text - which here becomes a tool for rendering digital gestures into a haptic format,
creating works that cannot be seen but touched. Accompanying each braille work is a digital counterpart
where source code preserves the exact instructions sent to the braille machines first developed in the
1980s.
The links below are real-time audiovisual digital work with machine drawing instructions, inscribed as
ordinals on BTC. (left-click for interaction)
All 20 inscriptions can be found here
Rooster 44
🐢Hawksbill 67
🐟Rooster 44
🐜Ant 35
⭐Rooster 44
🐘Elephant 19
🐙Octopus 27
🐆Leopard 111
🐈Wild Cat 5
🐊Crocodile 1
🐬Dolphin 47
🕊Pigeon 23
🦀Crab 19
🦋Butterfly 73
🦏Rhino 7
🦕Brontosaurus 17
🦘Kangaroo 33
🦙Llama 29
🦩Flamingo 59
🧸Panda 24
Process
