基于深度学习与双目立体视觉的物体管理应用

Application of object management based on deep learning and binocular stereovision

利用搭载双目摄像机与GPS设备,提出了一种基于深度学习与双目立体视觉的物体管理方案。首先用智能眼睛拍摄照片,并利用GPS获取智能眼睛所在的位置,然后利用卷积神经网络(CNN)中的FAST-RCNN对拍摄照片进行物体识别,获取照片中的物体,再利用双目立体视觉技术中的SGBM算法,获取照片中的物体相对与摄像机的坐标。利用拍摄者的GPS与物体相对拍摄者的坐标,就可以获取物体的坐标,从而获取物品的位置,实现管理物体的功能。

An object management scheme based on in-depth learning and binocular stereovision is proposed by using the equipment with binocular camera and GPS.Firstly, the smart eyes are used to take photos, and the position of the smart eyes is obtained by using GPS.Then, the FAST-RCNN in the convolution neural network (CNN) is used to recognize the object in the photograph and obtain the object in the photograph.Then, the SGBM algorithm in binocular stereovision technology is used to obtain the coordinates of the object in the photograph and the camera.Using the GPS of the photographer and the coordinates of the object relative to the photographer, the coordinates of the object can be obtained, thus realizing the function of managing the object.