2022 Fiscal Year Final Research Report
A Study on Automatic Estimation of Impression Attributes based on Human Visual Function
| Project/Area Number |
20K11864
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 61010:Perceptual information processing-related
|
|---|
| Research Institution | Tottori University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | 人物印象属性 / 自動推定 / 視覚特性 |
|---|
| Outline of Final Research Achievements |
We investigated the gaze time-series signals, which indicate where and when the observer pays attention to a region of an image when perceiving the subject as beautiful, elegant, or clean. We developed a method to show that the gaze location approaches the subject’s nose initially, and then the gaze location varies according to the impression words. Furthermore, we developed a new attention mechanism by introducing the gaze time-series signals into a deep-learning model and making it closer to human perception. We confirmed that the model improved the estimation accuracy.
|
| Free Research Field |
画像認識
|
| Academic Significance and Societal Importance of the Research Achievements |
システムが自動で人間の印象を推定できるようになると，その印象を手掛かりとしたコーディネートサービスを新たに創造できると考える．冠婚葬祭など非日常の特別な場で印象は重要となる．例えば，知人の結婚式へ出席するため，相応しい格好が与える印象を考えている．開発した推定方法を導入したシステムを構築できれば，これまでにないユーザ体験の提供が期待できる．その状況に適した格好を利用者は安心して選ぶことができ，自分では気付いていない格好を知る機会を得ることができる．
|
