利用深度学习估算托盘方位坐标

Estimation of Pallet Orientation Coordinates by Deep Learning

在工程机械领域,应用信息通信技术和人工智能等技术的工程机械数量在不断增加。因此,利用深度学习来识别从托盘边界框中提取的像素坐标,并利用像素尺寸变化与托盘边界框坐标之间的比例关系,通过摄像头深度识别的托盘孔来估算托盘坐标,其结果表明托盘的相对坐标估算可以达到要求的精度。

In the field of construction machinery,it is necessary to apply intelligent technologies such as Informa-tion Communication technology and Artificial Intelligence,and the number of construction machinery using this technology is increasing.Therefore,it uses Deep Learning to identify the pixel coordinates extracted from the pallet bounding box and uses the method of estimating pallet coordinates using the proportional relationship between pixel size changes and pallet bounding box coordinates,as well as the tray holes identified by camera depth recogni-tion,the results show that the relative coordinate estimation of the pallet can be achieved with the required accu-racy.