Palette Plate
Designed and coded an experimental kitchen tool that combined a Raspberry Pi, camera, projector, and AI to turn a countertop into a guide for creative plating.
Bringing digital guidance into the physical kitchen
Palette Plate began with a simple question: could technology help people compose a plate without pulling them away from the act of cooking? Our three-person design team imagined one integrated tool that could see the available ingredients and guide arrangement directly on the counter.
A camera captured the food and plate, a light maintained consistent visibility, and a projector returned the interface to the same physical surface. I owned the coding that connected those parts into one experience.
From ingredients to a projected composition
Computer vision identified the size and shape of the ingredients. Foodvisor and YOLOv8 contributed food recognition, GPT-3.5 interpreted the context, and image generation produced visual plating directions.
The system then projected the suggestion onto the counter as placement boundaries. Instead of translating instructions from a separate screen, users could arrange each ingredient inside a guide at the same scale and location as the final dish.
Prototyping the hardware and the wait
The physical components had to work together inside a compact form. We tested layouts and materials repeatedly to align the camera and projection while keeping the device understandable and practical around food.
Recognition and generation could take more than ten seconds, creating an awkward pause in the projected experience. I added loading motion and visible progress so the system acknowledged the wait instead of appearing broken.
Learning enough code to make the idea real
I started without a coding background and had to learn across unfamiliar hardware, computer vision, and AI services while building. Every integration exposed a new dependency, and every prototype made the gap between the concept and the physical experience more concrete.
The working prototype showed me the value of building through uncertainty. Coding was not separate from the design process—it was how I discovered timing constraints, hardware tradeoffs, and the interaction details that drawings alone could not reveal.