Food-101 - Mining Discriminative Components with Random Forests

  title={Food-101 - Mining Discriminative Components with Random Forests},
  author={Lukas Bossard and Matthieu Guillaumin and Luc Van Gool},
In this paper we address the problem of automatically recognizing pictured dishes. [] Key Result On the challenging mit-Indoor dataset, our method compares nicely to other s-o-a component-based classification methods.

Figures and Tables from this paper

BTBUFood-60: Dataset for Object Detection in Food Field

A novel dataset that includes images of 60 objects categories which are common in food domain, which is compared to PASCAL, Food-101, and ImageNet and baseline performance analysis for bounding-box detection results using a Faster Region based Convolutional Neural Networks is provided.

Highly Accurate Food/Non-Food Image Classification Based on a Deep Convolutional Neural Network

It is shown that CNN, which is the state of the art technique for general object classification, can perform accurately for this problem and achieve high accuracy 96, 95 and 99% in the three datasets respectively.

Food Image Classification with Deep Features

In this paper, deep feature extraction, feature concatenation and support vector machine (SVM) classifier are used for efficient classification of food images and it is seen that this method's performance is better than the other methods on FOOD-11 andFOOD-101 datasets.

Adapting New Categories for Food Recognition with Deep Representation

  • Shuang AoC. Ling
  • Computer Science
    2015 IEEE International Conference on Data Mining Workshop (ICDMW)
  • 2015
A new method that can alleviate the dataset bias for food image recognition by employing an external classifier for adaptation, called "negative classifier", which can learn efficient classifiers with fewer labeled examples.

An Efficient Food Image Classification By Inception- V3 Based Cnns

This paper adopted a Google Inception-V3 based convolutional neural networks (CNNs) model in order to classify food images and measured 92.89% of accuracy with considering the yummly and own dataset.

Food Image Recognition Using Very Deep Convolutional Networks

The effectiveness in classifying food images of a deep-learning approach based on the specifications of Google's image recognition architecture Inception appears to be at least closer to the requirements for mobile systems.

Visual Aware Hierarchy Based Food Recognition

A new two-step food recognition system that includes food localization and hierarchical food classification using Convolutional Neural Networks (CNNs) as the backbone architecture is proposed and Experimental results demonstrate that the system can significantly improve both classification and recognition performance on 4 publicly available datasets and the new VFN dataset.

A Deep Convolutional Neural Network for Food Detection and Recognition

  • M. SubhiS. M. Md Ali
  • Computer Science
    2018 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES)
  • 2018
A new deep convolutional neural network configuration is proposed to detect and recognize local food images and it was found out that convolution masks show that the features of food color dominate the features map.



Automatic Chinese food identification and quantity estimation

The issues of feature descriptors in the food identification problem are addressed, a preliminary approach for the quantity estimation using depth information is introduced and a straight forward method is proposed for certain food, while transparent food ingredients such as pure water and cooked rice are temporarily excluded.

PFID: Pittsburgh fast-food image dataset

The first visual dataset of fast foods is introduced with a total of 4,545 still images, 606 stereo pairs, 303 360° videos for structure from motion, and 27 privacy-preserving videos of eating events of volunteers to stimulate research on fast food recognition for dietary assessment.

Image Recognition of 85 Food Categories by Feature Fusion

This paper proposes an automatic food image recognition system for 85 food categories by fusing various kinds of image features including bag-of-features, color histogram, Gabor features and gradient histogram with Multiple Kernel Learning~(MKL).

Multiple-food recognition considering co-occurrence employing manifold ranking

  • Yuji MatsudaKeiji Yanai
  • Computer Science
    Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012)
  • 2012
It has been proved that co-occurrence statistics obtained from a food photo database is very helpful to improve the classification rate within the top ten candidates.

Randomized Clustering Forests for Image Classification

This work introduces Extremely Randomized Clustering Forests-ensembles of randomly created clustering trees-and shows that they provide more accurate results, much faster training and testing, and good resistance to background clutter.

Image Classification using Random Forests and Ferns

It is shown that selecting the ROI adds about 5% to the performance and, together with the other improvements, the result is about a 10% improvement over the state of the art for Caltech-256.

Combining randomization and discrimination for fine-grained image categorization

Results show that the proposed random forest with discriminative decision trees algorithm identifies semantically meaningful visual information and outperforms state-of-the-art algorithms on various datasets.

A food image recognition system with Multiple Kernel Learning

This paper proposes an automatic food image recognition system for recording people's eating habits and uses the Multiple Kernel Learning (MKL) method to integrate several kinds of image features such as color, texture and SIFT adaptively.

Blocks That Shout: Distinctive Parts for Scene Classification

This paper proposes a simple, efficient, and effective method to learn parts incrementally, starting from a single part occurrence with an Exemplar SVM, and can learn parts which are significantly more informative and for a fraction of the cost, compared to previous part-learning methods.

Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation

This paper proposes a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30% relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3%.