Symbody

At its core, Symbody functions as an intermediary between human movement and machine perception. The project uses advanced AI technology called auto-encoders, which work like sophisticated pattern recognition systems, to capture movement qualities in ways that traditional motion capture cannot. Rather than just tracking where a body moves in space, Symbody aims to understand and visualize how it moves – revealing patterns and expressions that might escape human perception.

Through the development process, several key technical components emerged: a customizable machine learning pipeline, a model database visualization tool, and a prototype for real-time interactive installation. These tools serve as critical intermediaries between raw movement data, machine processing, and human perception. The resulting prototype software allows users to explore latent space representations of movement data, enabling new ways of analyzing and understanding human motion. The system processes movement data in real-time, creating immediate visual representations that reveal hidden qualities of motion. These visualizations offer a new perspective for understanding human physicality and expression through the lens of artificial intelligence.

The project’s current prototype phase shows promising potential for various fields. In healthcare, it could contribute to early detection of movement disorders. Sports professionals might use it for detailed performance analysis, while choreographers and dancers could explore new dimensions of movement creation.

Through collaboration with TMC, In4Art, HEKA Lab, and the Institute of Artificial Intelligence Serbia, Symbody has evolved from an artistic concept into an experimental platform that bridges human intuition and machine perception. The project demonstrates how artificial intelligence can enhance our understanding of human expression, suggesting new possibilities for both artistic exploration and practical applications in movement analysis.