HOLD THE PLANTS
MAKE THEM DANCE
Plants had electrodes attached to turn them into touch sensors. AI images were produced by GANs trained on nature illustrations sourced from the Biodiversity Heritage Library, with different models for western plants, Japanese nature, and birds.
Algoflora models were trained using Projected GANs with a modified codebase to allow models to run at 40 FPS on a Macbook using CoreML. A GUI was also created to allow control of different generation parameters and debug hardware issues in real-time. Electrode values were read using an ESP32-S2 running CircuitPython.