Machine Learning Based Approach on Food Recognition and Nutrition Estimation

Zhidong Shen,Adnan Shehzad,Si Chen,Hui Sun,Jin Liu

DOI: https://doi.org/10.1016/j.procs.2020.06.113

2020-01-01

Procedia Computer Science

Abstract:Nowadays, standard intake of healthy food is necessary for keeping a balanced diet to avoid obesity in the human body. In this paper, we present a novel system based on machine learning that automatically performs accurate classification of food images and estimates food attributes. This paper proposes a deep learning model consisting of a convolutional neural network that classifies food into specific categories in the training part of the prototype system. The main purpose of the proposed method is to improve the accuracy of the pre-training model. The paper designs a prototype system based on the client server model. The client sends an image detection request and processes it on the server side. The prototype system is designed with three main software components, including a pre-trained CNN model training module for classification purposes, a text data training module for attribute estimation models, and a server-side module. We experimented with a variety of food categories, each containing thousands of images, and through machine learning training to achieve higher classification accuracy.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: to automatically classify food images and estimate nutrient components through machine - learning techniques, so as to help people better manage and monitor their daily dietary intakes, thereby avoiding obesity and related health problems.

Specifically, the paper points out:

1. **The severity of the obesity problem**: The adult obesity rate is rising sharply. High - calorie intakes can not only lead to obesity, but may also trigger a variety of diseases, such as breast cancer, colon cancer and prostate cancer, etc. According to the data of the World Health Organization, more than one - tenth of the world's adult population is in an obese state. In 2013, the American Medical Association officially defined obesity as a disease requiring medical intervention.

2. **Limitations of traditional methods**: Traditional dietary recording methods (such as food intake records in the past 24 hours) are effective to a certain extent, but are easily forgotten or rejected by users, resulting in inaccurate data or inability to continuously record.

3. **The proposed new method**: In order to overcome the shortcomings of traditional methods, researchers have proposed a new system based on deep learning. This system uses convolutional neural networks (CNN) to classify food images and estimate their nutrient components. The purpose of the system is to improve the accuracy of pre - trained models, so as to more accurately identify food categories and estimate their nutritional values.

In summary, this paper aims to develop an efficient and accurate food recognition and nutrition estimation system to help people better manage their diets, maintain a healthy weight, and prevent chronic diseases related to obesity.