-
Digica was tasked with creating a proof-of-concept gesture recognition system using three FMCW radar sources on prototype hardware for potential consumer applications. The project aimed to demonstrate the feasibility of gesture detection, differentiate between gestures and other movements, and achieve an end-to-end implementation including data acquisition, labelling, scenario creation, system development, model training, and precision evaluation. The system successfully achieved an 86% accuracy in gesture classification, detecting various gestures such as swipe, flick, approach, retract, and tap.
-
A leading producer of computer vision systems sought to explore the potential of migrating their image processing software onto a new heavily parallelized processor architecture. This transition aimed to provide a more configurable alternative to their existing FPGA implementation, without compromising performance. Ultimately, this effort aimed to offer a more open and customizable solution for their end-customers. The project successfully designed and implemented a high-level parallel processing software architecture, showcasing its suitability as a complementary option to FPGA with equivalent customizable functions.