Optical agri-food sorter uses AI close to the sensors

Agri-food sorting and harvesting machines benefit from compact AI cameras. The AI camera provides real-time segmented data: what do I see where. Easics’ AI solution interfaces directly with low-cost image sensors and doesn’t rely on the cloud. It contains an FPGA running a highly efficient deep neural network performing image interpretation. Orchid sorting is shown as a use case: from training till live demo.

About the speaker

Ramses Valvekens is managing director at Easics, a SoC, FPGA and ASIC design company, active in imaging, industry 4.0, healthcare, space and other markets. To this day, he vividly remembers his first encounter with neural networks, in 1996, when he used them to implement an optical character recognition system. In 1993 Valvekens co-invented, at Imec, the first soft microprocessor on FPGA, for which he received the Barco/VIK-prize. He holds two master degrees in electronics engineering from KU Leuven. He performed research at INP Grenoble (France) and at Lawrence Livermore National Laboratory (California).