• “CFLOW-AD: Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flows” has been accepted to WACV’22
  • “Contrastive Neural Processes for Self-Supervised Learning” has been accepted to ACML’21 (oral)


Kazuki Kozuka, Ph.D (小塚 和紀)

  • Contact: kozuka.kazuki [at]
  • Organization: Digital & AI Technology Center, Technology Division, Panasonic
  • Research: Image Recognition / Computer Vision

Kazuki Kozuka is a senior researcher at Panasonic Corporation. He received his Ph.D degrees from Nagoya Institite of Technology in 2009. In 2009, he joined Advanced Technology Laboratories, Panasonic Corporation, Japan, where he was involved in research on neuroscience, image recognition. He was also a visiting scholar in Stanford University between 2016 and 2019. His research interests lie in visual understanding through machine learning, mainly for computer vision and natural language processing. He received JCMI Best Presentation Award in 2013, JCMI Young Researcher Award in 2014, and MVA Best Poster Awards in 2015.


He is a researcher who specializes in collaboration, and has experienced joint research projects with various research institutes.


Ph.D in Engineering

2006 - 2009

Nagoya Institute of Technology, Nagoya, Japan

Adviser: Prof. Jun Sato

  • Research of multi-view geometry under more general conditions when using various types of sensors (1, 2, 3D, moving sensors, ...).
  • Multiple View Geometry of Mixed Dimensional Cameras.

M.S in Engineering

2004 - 2006

Nagoya Institute of Technology, Nagoya, Japan

Adviser: Prof. Jun Sato

  • Development of 3D shape reconstruction algorithm for auto-parking system.
  • Development of stabilization algorithm for feature extraction in translational camera and shape reconstruction by epipolar geometry.

Professional Experience

Senior Researcher

2021 - Present

Panasonic Corporation, Osaka, Japan

  • Lead projects to improve the amount/quality of data for AI and to develop it efficiently.
  • Lead AI research collaborations with top universities (Stanford University, U.C.Berkeley).

Visiting Scholar

2016 - 2019

Stanford University, California, USA

Adviser: Prof. FeiFei Li, Prof. Silvio Savarese, Dr. Juan Carlos Niebles

  • Lead a project on risk prediction technology for self-driving.

Chief Researcher

2012 - 2021

Panasonic Corporation, Osaka, Japan

  • Launched the Home Action Genome project to build/publish the world's largest action recognition dataset with Stanford University.
  • Launched the risk pediction project and started collaboration with Stanford University.
  • Launched an AI-based proteomics analysis project and started collaboration with Tokyo Institute of Technology.
  • Launching a collaboration with Yamaguchi University for diagnostic imaging support.
  • Developed a similar case retrieval system in collaboration with doctors at the University of Fukui.


2009 - 2012

Panasonic Corporation, Osaka, Japan

  • Developed a technology to search for similar cases by reproducing the points that doctors focus on during diagnosis.
  • Launched a project to support doctor's diagnosis, and starting a collaboration with the University of Fukui.
  • Developed 3D shape recognition for products and collaboration with the University of Tokyo.
  • Developed a technology to estimate sound comfort using ERP.

Collaboration Experience

Berkeley Artificial Intelligence Research Lab, U.C. Berkeley

2021 - present

Adviser: Prof. Trevor Darrell, Prof. Pieter Abbeel, Prof. Kurt Keutzer, Prof. Alexei Efros, Prof. Jitendra Malik

  • Research not disclosed.

Stanford Vision and Learning Lab, Stanford University

2016 - present

Adviser: Prof. FeiFei Li, Prof. Silvio Savarese, Dr. Juan Carlos Niebles, Prof. Ehsan Adeli

Machine Perception and Robotics Group, Chubu University

2015 - present

Adviser: Prof. Hironobu Fujiyoshi, Prof. Takayoshi Yamashita

  • Collaboration in the development of human trajectory prediction technology based on human regions and surrounding environment information.

Department of Community Health Care Promotion, University of Fukui

2021 - present

Adviser: Prof. Osamu Yamamura

  • Research not disclosed.

Computer Vision Laboratory, Tohoku University

2015 - 2016

Adviser: Prof. Takayuki Okatani

  • Research not disclosed.

Biological Physics Laboratory, Tokyo Institute of Technology

2014 - 2015

Adviser: Prof. Nobuhiro Hayashi

  • Collaboration in developing technology to predict sepsis using 2D electrophoresis images

Graduate School of Medicine, Yamaguchi University

2014 - 2015

Adviser: Prof. Shoji Kido

  • Collaboration in developing a diagnostic support system for lung CT.

School of Medical Sciences, University of Fukui

2011 - 2015

Adviser: Prof. Hirohiko Kimura, Prof. Toyohiko Sakai

  • Collaboration in the development of diagnostic support systems for lung CT (similar case retrieval system) and X-ray angiography.

JSK Laboratory, the University of Tokyo

2009 - 2012

Adviser: Prof. Masayuki Inaba

  • Collaboration in the development of classification technology for 3D data including deformed objects based on the similarity of overall shape and local shape distribution.

Cognitive Psychophysiology Laboratory, Hiroshima University

2009 - 2010

Adviser: Prof. Hiroshi Nittono

  • Collaboration in the development of automatic adjustment technology for hearing aids using event-related potentials.