Dish identification method

The invention provides a dish identification method. The method comprises the following steps of 1) obtaining a web request, and obtaining a corresponding image by a server according to the corresponding web request; 2) storing the image, obtaining an input data stream, generating an image file name and storing the image file name in a disk; 3) preprocessing the image, and performing size adjustment and normalization on the input image; and 4) performing processing by using a pre-trained convolutional neural network, detecting and classifying objects on the image, if dishes are not detected, ending the process, and if the dishes are detected, outputting corresponding dish information in combination with a classification result. According to the method, the defects in the prior art are overcome and a more accurate result can be output in real time. 本发明提供种菜品识别方法，包括以下步骤：1）获得web请求，服务器相应web请求，获取相应图像；2）保存图像，获取输入数据流，生成图像文件名并保存至磁盘；3）图像预处理，对输入的图像进行尺寸调整和归化；4）使用预先训练的卷积神经网络进行处理，对图像上的物体进行检测及分类，如果没有检测到菜品则结束，如果检测到菜品，则结合分类结果