Abstract

As a fine-grained classification problem, food image classification faces many difficulties in the specific implementation. Different countries and regions have different eating habits. In particular, Asian food images have a complicated structure, and the related classification methods are still very scarce. There is an urgent need to develop a feature extraction and fusion scheme based on the characteristics of Asian food images. To solve the above problems, we proposed an image classification model SLGC (SURF-Local and Global Color) that combines image segmentation and feature fusion. By studying the unique structure of Asian foods, the color features of the images are merged into the representation vectors in the local and global dimensions, respectively, thereby further enhancing the effect of feature extraction. The experimental results show that the SLGC model can express the intrinsic characteristics of Asian food images more comprehensively and improve classification accuracy.

Highlights

  • With the improvement of living standards, people began to pursue a more scientific and healthy diet

  • UEC FOOD 100 is a food image dataset created by Yoshiyuki Kawano of the University of Electro-Communications, most of which are popular Japanese foods, which can fully reflect the structural characteristics of Asian food

  • The segmentation of the food image can improve the classification accuracy by about 4% based on the same feature extraction method, the accuracy of classification can be improved by about 5% after local feature fusion, and 3% after the global feature fusion

Read more

Summary

Introduction

With the improvement of living standards, people began to pursue a more scientific and healthy diet. Hongsheng He et al present an automatic food classification method, DietCam, which addresses the variation of food appearances [3]. Since the food images mostly show the table scene, so there are inevitably hidden objects such as tableware, condiments, tablecloths, which increases the complexity of the stage Such problems lead to a lack of clarity in the food subject, which hurts the extraction of features, and seriously affects the effect of image classification. The “global colour feature” of the original image is extracted and merged with the image representation vector, input the merged features into a classification model based on the SVM (Support Vector Machine) for training and classification

The Theoretical Model
Feature Fusion
Bag of Features Model
Local Feature Fusion
Global Feature Fusion
Experimental Results and Analysis
Dataset
Image Segmentation
Experimental Results
Extraction methods
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call