Presentation 2019-05-31
Recognizing Fine-Grained Contexts at Home Using Multiple Cognitive APIs
Sinan Chen, Sachio Saiki, Masahide Nakamura,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Recognizing fine-grained contexts in individual houses and residents has been an important but challenging research topic. It has been studied for many years in the field of ubiquitous computing. We are studying fine-grained context recognition affordable for general households, by integrating general-purpose image-based cognitive APIs and light-weight machine learning. In our previous study, we first evaluated the recognition performance of commercial APIs, and then developed a recognition method that uses the recognition results of the single API as feature values. However, we found that the method could not distinguish different contexts with multiple people such as ``Eating'' and ``Playing games'' and ``General meeting''. The goal of this research is to improve the recognition accuracy of such difficult contexts, with preserving the affordability to general households. In the proposed method, we use multiple image-based cognitive APIs. For each API, we construct an independent recognition model using feature values of the API. Then, the context is determined by majority voting among results of the independent models. Compared to the popular approach with deep learning, the proposed method does not require a huge amount of labeled data or vast computing resources. As a result, it can recognize home contexts more accurately and economically.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Home context / Image recognition / Cognitive API / Machine learning / Majority voting
Paper # SC2019-6
Date of Issue 2019-05-24 (SC)

Conference Information
Committee SC
Conference Date 2019/5/31(2days)
Place (in Japanese) (See Japanese page)
Place (in English) National Institute for Materials Science
Topics (in Japanese) (See Japanese page)
Topics (in English) Science Service Platform, Data Service and Machine Learning, etc
Chair Masahide Nakamura(Kobe Univ.)
Vice Chair Shinji Kikuchi(National Institute for Materials Science) / Yoji Yamato(NTT)
Secretary Shinji Kikuchi(Tokyo University of Technology) / Yoji Yamato(Fujitsu Lab.)

Paper Information
Registration To Technical Committee on Service Computing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Recognizing Fine-Grained Contexts at Home Using Multiple Cognitive APIs
Sub Title (in English) Majority Voting Approach
Keyword(1) Home context
Keyword(2) Image recognition
Keyword(3) Cognitive API
Keyword(4) Machine learning
Keyword(5) Majority voting
1st Author's Name Sinan Chen
1st Author's Affiliation Kobe University(Kobe Univ.)
2nd Author's Name Sachio Saiki
2nd Author's Affiliation Kobe University(Kobe Univ.)
3rd Author's Name Masahide Nakamura
3rd Author's Affiliation Kobe University(Kobe Univ.)
Date 2019-05-31
Paper # SC2019-6
Volume (vol) vol.119
Number (no) SC-66
Page pp.pp.33-38(SC),
#Pages 6
Date of Issue 2019-05-24 (SC)