Food/Non-food Image Classification and Food Categorization using Pre-Trained GoogLeNet Model

Abstract

Recent past has seen a lot of developments in the field of image-based dietary assessment. Food image classification and recognition are crucial steps for dietary assessment. In the last couple of years, advancements in the deep learning and convolutional neural networks proved to be a boon for the image classification and recognition tasks, specifically for food recognition because of the wide variety of food items. In this paper, we report experiments on food/non-food classification and food recognition using a GoogLeNet model based on deep convolutional neural network. The experiments were conducted on two image datasets created by our own, where the images were collected from existing image datasets, social media, and imaging devices such as smart phone and wearable cameras. Experimental results show a high accuracy of 99.2% on the food/non-food classification and 83.6% on the food category recognition.

Grab, Pay, and Eat: Semantic Food Detection for Smart Restaurants

“…An improvement of about 4% is achieved in terms of overall accuracy using a method based on CNN [14]. From this, numerous researchers have proposed models based on CNNs either for feature extraction [2], [3] or for the whole recognition process [1], [4]. The best results obtained on public datasets with more than 15.000 images [1], [2] have been reported in [3] through the combination of CNN GoogLeNet for feature extraction, PCA for dimensionality reduction and SVM for classification.…”

“…The latter is quite relevant because remove the need for a manual selection of the chosen dishes, allowing to speed-up the service offered by these restaurants. From the computer vision side, several approaches have been proposed in order to tackle the problem, most of them using Convolutional Neural Networks (CNNs) [1], [2], [3], [4]. Several of the published work consider the development of methods for food recognition, that is, being able to recognize the dish depicted in a picture in which a single plate is shown.…”

MMHealth 2017

“…the authors were and are highly involved into several multimedia related international events in the field of the workshop: Measuring Behaviour 2016, PervasiveHealth 2014, SenseCam Symposium 2010, Beyond QuantifiedSelf at CHI 2014, among others. The organizers are working in the different fields of multimedia in health such as tracking of nutrition [1], digital interventions for personal health [2], activity recognition from Smart phones [4] and in lifelogging [3] using statistical methods, from multimodal heterogeneous data [4] and interpretation for personal health [6].…”

A Survey on Food Computing

“…Table 5 shows an overview of current performance comparison on benchmark datasets. [Anthimopoulos et al 2014] SIFT, Color -Food recognition [Oliveira et al 2014] Color, Texture -Mobile food recognition [Kawano and Yanai 2014c] HoG, Color -Mobile food recognition [Farinella et al 2015a] SIFT, Texture, Color -Food recognition [Martinel et al 2015] Color, Shape, Texture -Food recognition [Bettadapura et al 2015] SIFT, Color Location & Menu Restaurant-specific food recognition [Farinella et al 2015b] SIFT, SPIN -Food recognition SIFT, Color, HoG -Mobile food recognition [Ravl et al 2015] HoG, Texture, Color -Mobile food recognition [Martinel et al 2016] SIFT, Color, Shape, Texture -Food recognition [He et al 2017] Texture -Food recognition SIFT, Color -Food recognition Manuscript submitted to ACM Kawano and Yanai 2014b] HoG, Color, CNN -Food recognition [Simonyan and Zisserman 2014] VGG -Food recognition [Kagaya et al 2014] AlexNet -Food recognition [Ao and Ling 2015] GoogleNet -Food recognition AlexNet -Food recognition [Christodoulidis et al 2015] CNN -Food recognition VGG Text Recipe recognition DeCAF Location Restaurant-specific food recognition DeCAF Location Restaurant-specific food recognition [Herruzo et al 2016] GoogleNet -Food recognition [Wang et al 2016] CNN Location Restaurant-specific food recognition [Singla et al 2016] GoogleNet -Food recognition [Ragusa et al 2016] AlexNet, VGG, NIN -Food recognition GoogleNet -Food recognition [Ciocca et al 2016] AlexNet -Food recognition [Liu et al 2016] Inception -Food recognition [Hassannejad et al 2016] Inception -Food recognition [Tanno et al 2016 [Martinel et al 2018] WISeR -Food recognition…”

“…Compared with hand-crafted features, an improvement is achieved via CNN based deep networks[Kagaya et al 2014]. Numerous researchers have proposed CNN based models either for feature extraction[Aguilar et al 2017a;Ragusa et al 2016] or for the whole recognition process[Kagaya and Aizawa 2015;Singla et al 2016]. For example,[Singla et al 2016] reported the experiments on food/non-food classification using the GoogLeNet network.…”