EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Combining Deep Learning Mechanisms to Predict Interests from Gaze

Aoba Izumi, Fumiko Harada, Hiromitsu Shimakawa iD

Sensors · 09.05.2026 · 26 (10) · 2975

DOI: 10.3390/s26102975 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 09.05.2026. Verfasst von Aoba Izumi und 2 weiteren Autor:innen. Beteiligte Einrichtung: College of Information Science and Enginnering, Ritsumeikan University, Osaka 567-8570, Japan. Hauptthemen laut MEDLINE: Deep Learning, Fixation, Ocular. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper proposes a model to predict users’ personal interests from their gaze. Current image processing technologies enable us to identify each user’s gaze and pupil diameter. However, they cannot identify user interests. The model uses deep learning technologies. It consists of mechanisms to handle the two kinds of gaze: one specific to individuals and the other common to many people. Its training method utilizes the human property that user interests in objects in vision affect their gaze. Since it is known that pupil diameters increase when users view what they are interested in, we can train the model to predict users’ personal interests using the pupil diameters as labels. Notably, the paper not only proposes a model structure to predict personal interests but also examines the process of the modification of parameters of the model to reveal requisites to predict personal interests from gaze. The method enables us to provide personalized information to each user, such as recommendations in VoD services, advertisements during e-shopping, plate annotations in restaurant menus, and so on.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningFixation, OcularHumansImage Processing, Computer-AssistedPupil

Stichwörter der Autor:innen

deep learninggazeinterestsaliency maptransformer

Autor:innen und Institutionen

  • College of Information Science and Enginnering, Ritsumeikan University, Osaka 567-8570, Japan.

ORCID-Profile: Hiromitsu Shimakawa

Zitieren

APA

Izumi, A., Harada, F., & Shimakawa, H. (2026). Combining Deep Learning Mechanisms to Predict Interests from Gaze. Sensors, 26(10), 2975. https://doi.org/10.3390/s26102975

Vancouver

Izumi A, Harada F, Shimakawa H. Combining Deep Learning Mechanisms to Predict Interests from Gaze. Sensors. 2026;26(10):2975. doi:10.3390/s26102975

BibTeX
@article{Izumi2026,
  author = {Izumi, Aoba and Harada, Fumiko and Shimakawa, Hiromitsu},
  title = {Combining Deep Learning Mechanisms to Predict Interests from Gaze},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {10},
  pages = {2975},
  doi = {10.3390/s26102975}
}
Kostenlos starten →