3D Positioning Method for Pineapple Eyes Based on Multiangle Image Stereo-Matching

Currently, pineapple processing is a primarily manual task, with high labor costs and low operational efficiency. The ability to precisely detect locate eyes critical achieving automated eye removal. In this paper, machine vision automatic control technology are used build recognition positioning test platform, using the YOLOv5l target detection algorithm quickly identify images. A 3D localization based on multiangle image matching obtain position information of eyes, CNC precision motion system pierce probe into each verify effect algorithm. experimental results demonstrate that mAP reached 98%, average time required one was 0.015 s. According results, deviation between actual center penetration 1.01 mm, maximum 2.17 root mean square value 1.09 which meets accuracy requirements in eye-removal operations.