Real-Time Apple Detection System Using Embedded Systems With Hardware Accelerators: An Edge AI Application
This study shows the feasibility of deployment of the customized model on cheap and power-efficient embedded hardware without compromising mean average detection accuracy (83.64%) and achieved frame rate up to 30 fps even for the difficult scenarios such as overlapping apples, complex background, less exposure of apple due to leaves and branches.
Abstract
Real-time apple detection in orchards is one of the most effective ways of\nestimating apple yields, which helps in managing apple supplies more\neffectively. Traditional detection methods used highly computational machine\nlearning algorithms with intensive hardware set up, which are not suitable for\ninfield real-time apple detection due to their weight and power constraints. In\nthis study, a real-time embedded solution inspired from "Edge AI" is proposed\nfor apple detection with the implementation of YOLOv3-tiny algorithm on various\nembedded platforms such as Raspberry Pi 3 B+ in combination with Intel Movidius\nNeural Computing Stick (NCS), Nvidia's Jetson Nano and Jetson AGX Xavier. Data\nset for training were compiled using acquired images during field survey of\napple orchard situated in the north region of Italy, and images used for\ntesting were taken from widely used google data set by filtering out the images\ncontaining apples in different scenes to ensure the robustness of the\nalgorithm. The proposed study adapts YOLOv3-tiny architecture to detect small\nobjects. It shows the feasibility of deployment of the customized model on\ncheap and power-efficient embedded hardware without compromising mean average\ndetection accuracy (83.64%) and achieved frame rate up to 30 fps even for the\ndifficult scenarios such as overlapping apples, complex background, less\nexposure of apple due to leaves and branches. Furthermore, the proposed\nembedded solution can be deployed on the unmanned ground vehicles to detect,\ncount, and measure the size of the apples in real-time to help the farmers and\nagronomists in their decision making and management skills.\n