Eating Sound Dataset for 20 Food Types and Sound Classification Using Convolutional Neural Networks

Shijie, Jeannette; Maureira, Marcello A. Gómez; Rijn, Jan N. van

doi:10.1145/3395035.3425656

Cited by 4 publications

(3 citation statements)

References 14 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Audio features should not be neglected in investigating Computational Commensality, as eating is a highly multisensory experience (Spence, 2017). Fortunately, some data-sets are available in the literature focusing on audio features: the iHEARu-EAT database, for instance, features recordings from 30 subjects eating 6 different kinds of food (Hantke et al, 2016), whereas the Eating Sound data-set proposed by Ma et al (2020) includes audio from 20 different food types. Besides giving information on the kind of food that is consumed or on possible conversation topics, one could also exploit this modality to have a better picture of the commensal scenario.…”

Section: Related Workmentioning

confidence: 99%

What's on your plate? Collecting multimodal data to understand commensal behavior

et al. 2022

View full text Add to dashboard Cite

Eating is a fundamental part of human life and is, more than anything, a social activity. A new field, known as Computational Commensality has been created to computationally address various social aspects of food and eating. This paper illustrates a study on remote dining we conducted online in May 2021. To better understand this phenomenon, known as Digital Commensality, we recorded 11 pairs of friends sharing a meal online through a videoconferencing app. In the videos, participants consume a plate of pasta while chatting with a friend or a family member. After the remote dinner, participants were asked to fill in the Digital Commensality questionnaire, a validated questionnaire assessing the effects of remote commensal experiences, and provide their opinions on the shortcomings of currently available technologies. Besides presenting the study, the paper introduces the first Digital Commensality Data-set, containing videos, facial landmarks, quantitative and qualitative responses. After surveying multimodal data-sets and corpora that we could exploit to understand commensal behavior, we comment on the feasibility of using remote meals as a source to build data-sets to investigate commensal behavior. Finally, we explore possible future research directions emerging from our results.

show abstract

Section: Related Workmentioning

confidence: 99%

What's on your plate? Collecting multimodal data to understand commensal behavior

et al. 2022

View full text Add to dashboard Cite

show abstract

“…They developed a simple fully connected deep neural network to classify eating sounds of 20 different types of food. (Ma et al, 2020) The best model won using the MFCC as the input, which brought a 90% accuracy.…”

Section: Introductionmentioning

confidence: 99%

Classification of crispness of food materials by deep neural networks

Lopes,

Dacanal

2023

Journal of Texture Studies

View full text Add to dashboard Cite

Crispness is a textural characteristic that influences consumer choices, requiring a comprehensive understanding for product customization. Previous studies employing neural networks focused on acquiring audio through mechanical crushing of crispy samples. This research investigates the representation of crispy sound in time intervals and frequency domains, identifying key parameters to distinguish different foods. Two machine learning architectures, multi‐layer perceptron (MLP) and residual neural network (ResNet), were used to analyze mel frequency cepstral coefficients (MFCC) and discrete Fourier transform (DFT) data, respectively. The models achieved over 95% accuracy “in‐sample” successfully classifying fried chicken, potato chips, and toast using randomly extracted audio from ASMR videos. The MLP (MFCC) model demonstrated superior robustness compared to ResNet and predicted external inputs, such as freshly toasted bread acquired by a microphone or ASMR audio of toast in milk. In contrast, the ResNet model proved to be more responsive to variations in DFT spectrum and unable to predict the similarity of external audio sources, making it useful for classifying pretrained “in‐samples”. These findings are useful for classifying crispness among individual food sources. Additionally, the study explores the promising utilization of ASMR audio from Internet platforms to pretrain artificial neural network models, expanding the dataset for investigating the texture of crispy foods.

show abstract

“…They developed a simple fully connected deep neural network to classify eating sounds of 20 different types of food. (Ma, Gómez Maureira, & van Rijn, 2020) The best model won using the MFCC as the input, which brought a 90% accuracy.…”

Section: Introductionmentioning

confidence: 99%

Evaluation of sensory crispness of dry crispy foods by convolutional neural networks

Lopes

View full text Add to dashboard Cite

Título: "Avaliação da crocância sensorial de alimentos crocantes secos por redes neurais convolucionais" Agências de fomento: O presente trabalho foi realizado com apoio da Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Código de Financiamento 001 e pela Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)processo 2021/05317-9. ACKNOWLEDGEMENTSWhen people ask what is my purpose, I state: "I want to make everyone in the world to be capable of seeing, listening, and speaking." You would think why I worked on a project that's different from what I want to achieve soon. I thank you reader for thinking this way, we are here to learn that changing our perspectives brings us to our dreams. Life isn't linear, neither dreams nor plans.The results of this work are not mine alone, but of every person who gave an opinion.We may go faster alone, but together we can reach unexpected heights. I thank Professor Gelson Andrade da Conceição for being more than a teacher, but a friend who guided my heart and mind toward what was best for me. It was hard to transform my fixed mindset into a growth mindset. You prepared me to overcome the most difficult barrier that was making me regress in life: myself. Victor Dias de Oliveira, my best friend, you were always backing me up when I needed it. A few words can't describe how grateful I am for having you by my side. You have built a marketable version of the crispness analyzer, it's an honor to work with you to bring our initiatives into reality. I thank Professor Gustavo César Dacanal for the opportunity to work on this dissertation which is opening new opportunities at LAFLUSP. This work complemented with praise the areas that I seek to develop as a professional: the developing of new products, neural networks, and programming logic.Vivian Lara, you have planted the seed of entrepreneurship in my mind and my heart, I don't doubt that this work one day will be a reference in the control quality laboratories around Brazil, just wait and you will see.Daniela Correa and Larissa Rodrigues, you two have accompanied me in the laboratory days, that big yet hollow place, it's because of you two that we could bring energy to that place. I thank my family for supporting me.

show abstract

Eating Sound Dataset for 20 Food Types and Sound Classification Using Convolutional Neural Networks

Cited by 4 publications

References 14 publications

What's on your plate? Collecting multimodal data to understand commensal behavior

What's on your plate? Collecting multimodal data to understand commensal behavior

Classification of crispness of food materials by deep neural networks

Evaluation of sensory crispness of dry crispy foods by convolutional neural networks

Contact Info

Product

Resources

About