Researchers

Masao Yamanaka

Department:

Solution Research Div.
Cognitive Data Engineering Grp.
Group Leader/Principal Researcher

Research Domains:

Research and development on video data processing and big-data analysis

Biography:

2015–present
Engaged in research and development based on statistical machine learning, including technology to detect the state of the passengers inside a vehicle (posture, behavior and level of alertness) by using data collected from sensors installed within the vehicle, technology to detect obstacles on the road using images taken from an onboard camera, technology to assess the workload of workers inside a plant using images, technology to guess the subject of sentences within conversations with entertainment robots, and technology to shed light on factors behind the decline in the yield rate of vehicle parts.

2007–2015
Worked in research and development on image data processing technologies at a major precision equipment manufacturer based on statistical machine learning, including personal authentication, human body detection, posture estimation, detection of abnormal behavior, estimation of the main subject in photographs, and aesthetic evaluation.
2001–2007
Involved in the development of digital cameras, camcorders and other large-scale integration (LSI) circuits at a major electronics manufacturer.

For some time, the term AI has been used to refer to a variety of applications based on statistical machine learning. However, the truth is that nothing can be called artificial intelligence in the real sense of the word. So far, AI has achieved a certain degree of results, only through humans using their true intelligence to supplement the AI's functional imperfections to make them think that it is perfect. In the past, the popularity of AI research and development has risen or fallen in regular cycles, but if we truly want to enrich our society as humans, we need research and development that brings AI closer to technologies that have a high affinity with it, rather than bringing AI closer to true intelligence. Based on this philosophy, I would like to aim to conduct a wide range of initiatives in research and development.

Biography

2015–present
Engaged in research and development based on statistical machine learning, including technology to detect the state of the passengers inside a vehicle (posture, behavior and level of alertness) by using data collected from sensors installed within the vehicle, technology to detect obstacles on the road using images taken from an onboard camera, technology to assess the workload of workers inside a plant using images, technology to guess the subject of sentences within conversations with entertainment robots, and technology to shed light on factors behind the decline in the yield rate of vehicle parts.

2007–2015
Worked in research and development on image data processing technologies at a major precision equipment manufacturer based on statistical machine learning, including personal authentication, human body detection, posture estimation, detection of abnormal behavior, estimation of the main subject in photographs, and aesthetic evaluation.
2001–2007
Involved in the development of digital cameras, camcorders and other large-scale integration (LSI) circuits at a major electronics manufacturer.

Publication List/Conference presentation

Paper
・Salient Region Detection by Enhancing Diversity of Multiple Priors. IPSJ Transactions on Mathematical Modeling and Its Applications Vol.9 No.1 13–22 (Feb. 2016).
・Driver's Behavior Recognition Based on the Global Architecture of Human Parts Position. The 22th Symposium on Sensing via Image Information, 2016.
・Detection of activities and events without explicit categorization. IPSJ Transactions on Mathematical Modeling and Its Applications, vol.6, no.2, pp.86-92, 2013.
・Salient object detection based on direct density-ratio estimation. IPSJ Transactions on Mathematical Modeling and Its Applications, vol.6, no.2, pp.78-85, 2013.

Conference presentation
An Symposium on AI and SOCIETY, 2017.10.10.tue-10.11.web