基于卷积神经网络在菜单识别的应用

Application of Convolutional Neural Network in Menu Recognition

菜单图像分割是进行菜单识别的首要步骤,分割的准确性会影响菜单识别的效果。该文对线下餐饮店菜单识别不准确问题进行研究,利用深度学习的卷积神经网络技术对餐饮店菜单进行识别。首先利用数字图像处理技术对拍照菜单图片进行预处理,其次通过卷积神经网络技术预处理后的图像进行识别。实现准确反馈给店家顾客具体的点餐信息,以实现服务人员直接上传点餐照片,便可以直接在本店系统上完成点餐的功能,提高了识别的准确性。

Menu image segmentation is the first step of menu recognition,and the accuracy of segmentation will affect the effect of menu recognition.This article studies the problem of inaccurate menu recognition in offline restaurants,and deep learning convolutional neural network technology was used to identify menu in restaurants.Firstly,digital image processing technology is used to preprocess the photo menu image,and then the pre-processed image is identified by convolutional neural network technology.Realizing accurate feedback to the store customers’specific order information,so that the service staff can directly upload the order photos,and then the ordering function can be completed directly on the store system,which improves the accuracy of recognition.