DearDiary
A space that writes its own memories through sensing and machine handwriting.
Object Recognition IBM Watson API
Web Crawler
SVG-to-Gcode Conversion
GRBL CNC
Sensor-Based Documentation
TIME | 2019
AUTHORS | Po-Yao Wu
EXHIBITIONS |
- 2019 International Exhibition of Electronic Art SYNTOPIA | Athens, Greece.
MOTIVATION |
DearDiary imagines what a space would write if it could keep a diary of its own. Instead of direct video recording, the project treats “writing” as an interpretive process—one that filters, transforms, and subjectively translates what is sensed. Using object recognition, DearDiary observes its surroundings and converts detected elements into symbolic icons, which are then handwritten onto paper by a CNC-based drawing machine. Over time, these machine-written diary pages become a temporal archive of the space’s daily impressions, capturing people, objects, and environmental changes as a sequence of subjective memories. The work reflects on how spaces accumulate meaning and how machine perception might reinterpret everyday life.
APPROACH |
The system consists of a camera for environmental sensing and a custom GRBL-driven writing machine that draws diary entries. Real-time object recognition—originally via IBM Watson API—identifies elements within the space. A custom web crawler retrieves corresponding icons from online libraries, which are converted from SVG to G-code and passed to the CNC plotter.
Each diary page contains roughly nine icons and takes about an hour to complete. The output is entirely mechanical handwriting, freezing digital observations into physical marks. As the machine senses its environment and writes continuously, viewers and surrounding objects become part of its diary content—turning the space into an active narrator.
