First Bite/Chew: distinguish typical allergic food by two IMUs

Juling Li, Xiongqi Wang, Junyu Chen, Thad Starner, George Chernyshov, Jing Huang, Yifei Huang, Kai Kunze, Qing Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Eating or overtaking allergic foods may cause fatal symptoms or even death for people with food allergies. Most current food intake tracking methods are camera-based, on-body sensor-based, microphone based, and self-reported. However, challenges that remain are allergic food detection, social acceptance, lightweight, easy to use, and inexpensive. Our approach leverages the first bite/chew and the corresponding hand movement as an indicator to distinguish typical types of the allergic food. Our initial feasibility study shows that our approach can distinguish six types of food at an accuracy of 89.7% over all four participants' mixed data. Particularly, our method successfully detected and distinguished typical allergic foods such as burgers (wheat), instant noodles (wheat), peanuts, egg fried rice, and edamame, which can be expected to contribute to not only personal use but also medical usage.

Original languageEnglish
Title of host publicationProceedings 4th Augmented Humans International Conference, AHs 2023
PublisherAssociation for Computing Machinery
Number of pages4
ISBN (Electronic)9781450399845
Publication statusPublished - 2023 Mar 12
Event4th Augmented Humans International Conference, AHs 2023 - Glasgow, United Kingdom
Duration: 2023 Mar 122023 Mar 14

Publication series

NameACM International Conference Proceeding Series


Conference4th Augmented Humans International Conference, AHs 2023
Country/TerritoryUnited Kingdom


  • diet monitoring
  • food intake
  • smart eyewear

ASJC Scopus subject areas

  • Software
  • Human-Computer Interaction
  • Computer Vision and Pattern Recognition
  • Computer Networks and Communications


Dive into the research topics of 'First Bite/Chew: distinguish typical allergic food by two IMUs'. Together they form a unique fingerprint.

Cite this