Preliminary results of peach detection in images applying convolutional neuronal network

The fruit detection part is very important for a good performance in a yield estimation system. This paper presents the preliminary results using the object detection Faster R-CNN method in the peaches images. The aim is evaluate the method performance in the detection of peach RGB images. Images acquired in an orchard were used. Although this method of object detection has been applied in other studies to detect fruits, according to the literature, it has not been used to detect peaches. The results, although preliminary, show a great potential of using the method to detect peach.
Este trabalho de investigação é financiado pelo projeto PrunusBot - Sistema robótico aéreo autónomo de pulverização controlada e previsão de produção frutícola, Operação n.º PDR2020-101-031358 (líder), Consórcio n.º 340, Iniciativa n.º 140, promovido pelo PDR2020 e co-financiado pelo FEADER e União Europeia no âmbito do Programa Portugal 2020.