Catering health analysis method based on image semantic deep learning

Abstract

The invention provides a catering health analysis method based on image semantic deep learning, which takes menu images as input and can achieve high-precision dish image classification and dish nutrient calculation. In the dish image classification part, a dish image classification network capable of learning the distance between menus is constructed, the network takes dish images and menu information as input, the information of raw material parts in the menus is deeply understood while the image information is learned, and the classification accuracy is further improved. In the nutrient calculation part, pixel-level semantic segmentation is carried out on an image, image information represented by each pixel point is accurate, the proportion of various raw materials in each image is clear, and the content of the raw materials in the menu is further corrected. For the same dish, different pictures can return different nutrient content information, so that the nutrient calculation module is more accurate and scientific.

Description

technical field [0001] The present invention mainly relates to technologies related to computer vision, and in particular to deep learning-based dish image recognition and dish image semantic segmentation technology. Background technique [0002] In today's society, healthy diet has become a topic of concern and concern to ordinary people. A sensible and healthy diet can also help people prevent diet-related diseases such as diabetes. However, at this stage, the science and popularization of healthy diet is still not enough, and most people still don't know enough about the real scientific diet health. Therefore, dietary health requires not only increased attention, but more importantly, a way to help the masses scientifically understand catering and give guidance and suggestions with medical value. What the masses need is not just a perceptual understanding of diet, Guidance on specific figures and data is more needed. [0003] In terms of dish analysis system, most of t...

AI Technical Summary

Problems solved by technology

[0003] In terms of dish analysis system, most of the related catering analysis systems at this stage have two defects: users need to have relevant knowledge of dishes, such as raw materials, recipes, etc.; the nutrient information provided is not comprehensive enough, and there is a lack of scientific medical guidance and suggestions

This type of method requires a lot of additional artificial information, such as the prior relationship between the ingredients of the dishes, etc., and there is no in-depth study of the recipe pictures, and there is still room for improvement in the accuracy rate.

[0007] At the same time, there is currently no pixel-level semantic segmentation of the raw materials of the dish pictures. For the data query of the dishes, the calculation of the nutrients of the dishes stays on the surface, that is, the nutrients per unit mass obtained from different pictures of the same kind of dishes are the same, and it is impossible to target Detailed analysis and calculation of user pictures

Images