Use of machine learning in the developed weeding machine (title)
When developing a new weeding machine enabling automatic recognition of crops from weeds, the company Ullmanna s.r.o. needed to use machine learning. The aim of the cooperation was to create and use the HPC infrastructure to verify the neural network model applicable to inference on the terminal device (weeding machine) while meeting the limiting conditions of deployment (limited HW, inference speed, recognition accuracy). The experiment is still running. With regard to its potential, the project “NEWMAN” (Non-chEmical Weeding MAchiNe) has been selected for TOP5 supported out of a total of 93 applications in the AgROBOfood.eu call, with DIH Ostrava providing a key service in the project.