IEICE Technical Committee Submission System
Advance Program
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top  Go Back   / [HTML] / [HTML(simple)] / [TEXT]  [Japanese] / [English] 

===============================================
Technical Committee on Pattern Recognition and Media Understanding (PRMU)
Chair: Seiichi Uchida (Kyushu Univ.) Vice Chair: Takuya Funatomi (NAIST), Mitsuru Anpai (Denso IT Lab.)
Secretary: Kouta Yamaguchi (CyberAgent), Yusuke Matsui (Univ. of Tokyo)
Assistant: Nakamasa Inoue (Tokyo Inst. of Tech.), Yasutomo Kawanishi (Riken)

===============================================
Technical Committee on Information-Based Induction Sciences and Machine Learning (IBISML)
Chair: Masashi Sugiyama (Univ. of Tokyo) Vice Chair: Toshihiro Kamishima (AIST), Koji Tsuda (Univ. of Tokyo)
Secretary: Tomoharu Iwata (NTT), Atsuyoshi Nakamura (Hokkaido Univ.)
Assistant: Yoshinobu Kawahara (Osaka Univ.), Taiji Suzuki (Tokyo Inst. of Tech.)

===============================================
Special Interest Group on Computer Vision and Image Media (IPSJ-CVIM)


DATE:
Thu, Mar 2, 2023 09:00 - 18:10
Fri, Mar 3, 2023 09:00 - 17:55

PLACE:


TOPICS:


----------------------------------------
Thu, Mar 2 AM (09:00 - 10:15)
----------------------------------------

(1)/PRMU 09:00 - 09:10
Toward Regularizing Neural Networks with Meta-Learning Generative Models
Shin'ya Yamaguchi (NTT/Kyoto Univ.), Daiki Chijiwa, Sekitoshi Kanai, Atsutoshi Kumagai (NTT), Hisashi Kashima (Kyoto Univ.)

(2)/PRMU 09:10 - 09:20
Adversarial Training Focusing on Misclassification Risk in Image Classification
Norikoe Murakami, Seiya Ito, Kouzou Ohara (Aoyama Gakuin Univ.)

(3)/PRMU 09:20 - 09:30
Relation Between Shape/Texture Biases and Double-Descent Phenomenon in Visual Recognition
Shuya Takahashi (Tokyo Denki Univ./AIST), Nakamasa Inoue, Rio Yokota (Tokyo Tech), Hirokatsu Kataoka (AIST), Eisaku Maeda (Tokyo Denki Univ.)

(4)/PRMU 09:30 - 09:40
Detection of UI parts from web page images utilizing a small amount of data
Wataru Aizaki, Sotaro Kato, Yoshihisa Shinozawa (Keio Univ.), Yukako Iimura, Shinobu Saito (NTT)

(5) 09:40 - 10:15


----------------------------------------
Thu, Mar 2 AM (09:00 - 10:15)
----------------------------------------

(6)/PRMU 09:00 - 09:10
A Study on Simultaneous Detection of Music Notes and Staff with YOLOv7
En Kamiya, Nobutaka Kuroki, Masahiro Numa (Kobe Univ.)

(7)/PRMU 09:10 - 09:20
LabOps: Automating Research and Education Activities
Yusuke Matsui (UT)

(8)/PRMU 09:20 - 09:30
Distributed Parallel Construction of Approximate Nearest Neighbor Search Index for Large-Scale and High-Dimensional Data
Naoki Ono, Yusuke Matsui (UTokyo)

(9)/PRMU 09:30 - 09:40
(See Japanese page.)

(10)/PRMU 09:40 - 09:50
Pruning for Producing Efficient DNNs: Neuron Selection and Reconstruction based on Final Layer Error
Koji Kamma, Toshikazu Wada (Wakayama Univ.)

(11) 09:50 - 10:15


----------------------------------------
Thu, Mar 2 AM (09:00 - 10:15)
----------------------------------------

(12) 09:00 - 09:40


(13) 09:40 - 10:15


----- Break ( 10 min. ) -----

----------------------------------------
Thu, Mar 2 AM (10:25 - 11:55)
----------------------------------------

(14)/PRMU 10:25 - 10:35
Cell mitosis detection with partial annotation
Kazuya Nishimura (Kyushu Univ.), Ami Katanaya, Shinichiro Chuma (Kyoto Univ.), Ryoma Bise (Kyushu Univ.)

(15)/PRMU 10:35 - 10:45
(See Japanese page.)

(16)/PRMU 10:45 - 10:55
Study on Estimation Method of Landslide Location and Scale by Segmentation of RGBD Images Acquired by the Camera Attached to a Drone
Kai Matsui (Waseda Univ.), Katsuya Hasegawa (JAXA), Jun Ohya, Yoichi Kato, Mako Yokozawa (Waseda Univ.)

(17)/PRMU 10:55 - 11:05
Investigation of a Method to Obtain Growing Conditions by Object Detection for Cucumber image under Forcing Culture Using IoP Cloud
Yuma Usui, Akihiro Tanikawa, Takaya Matsumoto (KUT/Office asoT), Mitsuhiro Osaki, Kazuma Nishiuchi (Citynet), Masahiro Fukumoto (KUT)

(18)/PRMU 11:05 - 11:15
On the Effectiveness of Formula-Driven Supervised Learning for Medical Image Tasks
Ryuto Endo, Shuya Takahashi, Eisaku Maeda (TDU)

(19)/PRMU 11:15 - 11:25
A trial of writer identification using a deep neural network trained by handwriting of multiple character types
-- A study using a handwriting database of 401 writers --
Yoshinori Akao (NRIPS)

(20) 11:25 - 11:55


----------------------------------------
Thu, Mar 2 AM (10:25 - 11:55)
----------------------------------------

(21)/PRMU 10:25 - 10:35
*
Masaya Mizuno (Nagoya Univ.), Tomohiro Fujita, Yasutomo Kawanishi (RIKEN), Daisuke Deguchi, Hiroshi Murase (Nagoya Univ.)

(22)/PRMU 10:35 - 10:45
Investigation of Appearance Inspection Method Considering the Number of Corresponding Local Patches
Katsuhisa Kitaguchi, Yohei Nishizaki, Mamoru Saito (ORIST)

(23)/PRMU 10:45 - 10:55
A Study on Trendy Style Learning and Recommendation in Digital Makeup System
Yoshimi Sato, Nobutaka Kuroki, Masahiro numa (Kobe Univ.)

(24)/PRMU 10:55 - 11:05
Report on the 3rd Lip-Reading Challenge
Takeshi Saitoh (Kyutech), Yuto Goto, Hiroyuki Nagano, Akihiro Kato, Masaki Nose (RICOH), Naoki Hiramoto (Mercoin), Tomohiro Hattori, Shiiya Aoyama, Yusuke Katoh, Ryuta Toshima, Takumi Nagawaki, Satoshi Tamura (Gifu University), Daiki Arakane (Kyutech)

(25)/PRMU 11:05 - 11:15
A Study of Word Lip-Reading using Meta Learning
Michinari Kodama, Takeshi Saitoh (kyutech)

(26)/PRMU 11:15 - 11:25
Dynamic Weight Scheduling for Long-tailed Visual Recognition
Xinyuan Li (Ritsumeikan Univ.), Yu Wang (Hitotsubashi Univ.), Jien Kato (Ritsumeikan Univ.)

(27) 11:25 - 11:55


----------------------------------------
Thu, Mar 2 AM (10:25 - 11:55)
----------------------------------------

(28)/IBISML 10:25 - 10:40
Blind deconvolution with non-smooth regularization via Bregman proximal DC algorithms
Shota Takahashi (SOKENDAI), Mirai Tanaka, Ikeda Shiro (ISM)

(29)/IBISML 10:40 - 10:55
Boosting for Class Imbalance Data
Yuya Saito, Shinnosuke Matsuo, Seiichi Uchida, Daiki Suehiro (Kyushu Univ)

(30)/IBISML 10:55 - 11:10
Non-uniqueness of solutions in Physics-informed neural networks
Shuichiro Tsuda, Toshiyuki Tanaka (Kyoto Univ.)

(31)/IBISML 11:10 - 11:25
Optimal Approximation Rate of Deep CNNs with Small Filter Size
Yuma Sato, Masaaki Imaizumi (UT)

(32)/IBISML 11:25 - 11:40
Binarization of Vision Transformer with Scaling Factors
Shun Sato, Shun Sawada, Hidefumi Ohmura, Kouichi katsurada (TUS)

(33)/IBISML 11:40 - 11:55
Novel Adversarial Attacks Based on Embedding Geometry of Data Manifolds
Masahiro Morita, Hajime Tasaki, Jinhui Chao (Chuo Univ.)

----- Break ( 95 min. ) -----

----------------------------------------
Thu, Mar 2 PM (13:30 - 14:20)
----------------------------------------

(34)/PRMU 13:30 - 14:20
[Fellow Memorial Lecture]
Team Building for World-Class CV x AI
-- Anticipating changes in the world and utilizing brain networks to flexibly and quickly put top technologies to practical use --
Masato Kawade (SenseTime Japan)

----- Break ( 10 min. ) -----

----------------------------------------
Thu, Mar 2 PM (14:30 - 16:30)
----------------------------------------

(35)/IBISML 14:30 - 15:10
[Invited Talk]
--
Masato Ishii (Sony)

(36)/IBISML 15:10 - 15:50
[Invited Talk]
--
Yuma Koizumi (Google Research)

(37)/IBISML 15:50 - 16:30
[Invited Talk]
--
Yusuke Tashiro (MTEC)

----- Break ( 10 min. ) -----

----------------------------------------
Thu, Mar 2 PM (16:40 - 18:10)
----------------------------------------

(38)/PRMU 16:40 - 16:50
Does the class imbalance in the pre-training always adversely affect transfer learning performance?
Shojun Nakayama (Toshiba)

(39)/PRMU 16:50 - 17:00
Evaluating importance of data quantity and domain diversity for training dataset
Eisho Tsuji, Shintatro Takenaga (Univ. Tsukuba/AIST), Masaki Onishi (AIST)

(40)/PRMU 17:00 - 17:10
Evaluation of Spatial Transformer Network with Vision Transformer
Yoshiharu Iwasaki (Ritsumeikan Univ.), Yu Wang (Hitotsubashi Univ.), Jien Kato (Ritsumeikan Univ.)

(41)/PRMU 17:10 - 17:20
learning from label proportions
Shinnosuke Matsuo, Daiki Suehiro, Seiichi Uchida, Ryoma Bise (Kyushu Univ.)

(42)/PRMU 17:20 - 17:30
Shape completion
Yusuke Nagata, Seiichi Uchida (Kyushu Univ.)

(43)/PRMU 17:30 - 17:40
A Study on Data Augmentation by Pixel Value Transformation for Product Image Retrieval DNN
Yu Okamoto, Masaki Kishibe, Toshikazu Wada (Wakayama Univ.)

(44) 17:40 - 18:10


----------------------------------------
Thu, Mar 2 PM (16:40 - 17:55)
----------------------------------------

(45)/PRMU 16:40 - 16:50
Separate Predictive Time Series Reconstruction using Independent Low-Rank Matrix Analysis
Hibiki Izumi, Jun Rokui (Univ of Shizuoka)

(46)/PRMU 16:50 - 17:00
(See Japanese page.)

(47)/PRMU 17:00 - 17:10
A Semi-Supervised Learning Framework for Handwritten Text Recognition using Mixed Augmentations and Scheduled Pseudo-Label Loss
Masayuki Honda, Hung Tuan Nguyen, Cuong Tuan Nguyen (TUAT), Cong Kha Nguyen, Ryosuke Odate, Takashi Kanemaru (Hitachi Ltd.), Masaki Nakagawa (TUAT)

(48) 17:10 - 17:55


----------------------------------------
Thu, Mar 2 PM (16:40 - 17:55)
----------------------------------------

(49) 16:40 - 17:30


(50) 17:30 - 17:55


----------------------------------------
Fri, Mar 3 AM (09:00 - 10:00)
----------------------------------------

(51)/PRMU 09:00 - 09:10
(See Japanese page.)

(52)/PRMU 09:10 - 09:20
Unknown Object Segmentation by View Independent Scene Change Detection
Li Jiaxin (Nagoya Univ.), Yasutomo Kawanishi (RIKEN), Daisuke Deguchi, Hiroshi Murase (Nagoya Univ.)

(53)/PRMU 09:20 - 09:30
3D Point Cloud based Object Recognition and Auto True-Value System Construction for Autonomous Driving
Bin Zhang, Congzhi Ren, Hun-ok Lim (KU)

(54) 09:30 - 10:00


----------------------------------------
Fri, Mar 3 AM (09:00 - 10:00)
----------------------------------------

(55) 09:00 - 09:20


(56)/PRMU 09:20 - 09:30
A Study of Few-shot NeRF by Pseudo-Feature Vectors Evaluation for Unknown Viewpoints
Daiju Kanaoka (Kyutech/RIKEN), Motoharu Sonogashira (RIKEN), Hakaru Tamukoh (Kyutech/Neumorph Center), Yasutomo Kawanishi (RIKEN)

(57)/PRMU 09:30 - 09:40
Memory Efficient Trilinear Up-sampling for Sparse Voxels
Riku Yamashita, Yusuke Matsui (UTokyo)

(58) 09:40 - 10:00


----------------------------------------
Fri, Mar 3 AM (09:00 - 10:00)
----------------------------------------

(59) 09:00 - 09:40


(60) 09:40 - 10:00


----- Break ( 5 min. ) -----

----------------------------------------
Fri, Mar 3 AM (10:05 - 11:05)
----------------------------------------

(61)/PRMU 10:05 - 10:15
High Accuracy Alignment of 3D Point Clouds Obtained from Multiple View Point Images for Automatic Wire Harness Placement Environments
Erika Hata, Eiji Fukuzawa, Jun Ohya (Waseda Univ.)

(62)/PRMU 10:15 - 10:25

Satoshi Kamiya (Meijo University), Youichi Sakurada, Kenji Kashiwagi (Yamanashi University), Kazuhiro Hotta (Meijo University)

(63)/PRMU 10:25 - 10:35
(See Japanese page.)

(64)/PRMU 10:35 - 10:45
(See Japanese page.)

(65) 10:45 - 11:05


----------------------------------------
Fri, Mar 3 AM (10:05 - 11:05)
----------------------------------------

(66) 10:05 - 10:25


(67)/PRMU 10:25 - 10:35
Consistent 3D Stylization in sketch with NeRF
Yanyuan Fu, Yusuke Matsui (UTokyo)

(68)/PRMU 10:35 - 10:45
A Basic Study on Automated Camera Selection for Vehicle Localization using Scene Recognition Features
Haruya Kyutoku (Aichi Univ. of Tech./Toyota Tech. Inst.), Tokihiko Akita, Seiichi Mita (Toyota Tech. Inst.)

(69) 10:45 - 11:05


----------------------------------------
Fri, Mar 3 AM (10:05 - 11:05)
----------------------------------------

(70) 10:05 - 10:25


(71)/PRMU 10:25 - 10:35
[Short Paper]
Local deformation feature extraction with Deformable Convolution
Kazuki Kitajima, Seiichi Uchida (Kyushu Univ.)

(72) 10:35 - 11:05


----- Break ( 10 min. ) -----

----------------------------------------
Fri, Mar 3 AM (11:15 - 12:05)
----------------------------------------

(73) 11:15 - 12:05


----- ( 60 min. ) -----

----------------------------------------
Fri, Mar 3 PM (13:05 - 13:50)
----------------------------------------

(74) 13:05 - 13:50


----- Break ( 10 min. ) -----

----------------------------------------
Fri, Mar 3 PM (14:00 - 16:00)
----------------------------------------

(75) 14:00 - 16:00


----- Break ( 10 min. ) -----

----------------------------------------
Fri, Mar 3 PM (16:10 - 17:40)
----------------------------------------

(76)/PRMU 16:10 - 16:20

Sota Kato, Kazuhiro Hotta (Meijo Univ.)

(77)/PRMU 16:20 - 16:30
Collaborative Intelligence for Transformer-Based AI Systems
Monikka Roslianna Busto, Shohei Enomoto, Takeharu Eda (NTT SIC)

(78)/PRMU 16:30 - 16:40
Eye-contact Transformer: Eye contact detection of distant pedestrians using a sequence of skeletons and scene features
Ryusei Hata, Daisuke Deguchi (Nagoya Univ), Takatsugu Hirayama (UHE/Nagoya Univ), Yasutomo Kawanishi (RIKEN/Nagoya Univ), Hiroshi Murase (Nagoya Univ)

(79)/PRMU 16:40 - 16:50
(See Japanese page.)

(80)/PRMU 16:50 - 17:00
Parallel-Data-Free Japanese Singer Conversion using CycleGAN Considering Perceptual Loss in Singing Phoneme Sequences
Kanade Gemmoto, Nobutaka Shimada, Tadashi Matsuo (Ritsumeikan Univ)

(81)/PRMU 17:00 - 17:10
Explainable Deep Clustering for Wafer Defect Pattern Classification
Yuki Okazaki, Hiroki Takahashi (The Univ. of Electro-Communications)

(82) 17:10 - 17:40


----------------------------------------
Fri, Mar 3 PM (16:10 - 17:40)
----------------------------------------

(83)/PRMU 16:10 - 16:20
Analysis of TV commercial favorability by scene labeling and XAI
Kanta Fukuyori, Kunio Matsui (Knazawa Inst.Tech.), Koh Tatsumoto (TOKYO KIKAKU CO.,Ltd.)

(84)/PRMU 16:20 - 16:30
Classifying Cable Tendency with Semantic Segmentation by Utilizing Real and Simulated RGB Data
Pei-Chun Chien, Powei Liao, Eiji Fukuzawa, Jun Ohya (Waseda Univ.)

(85)/PRMU 16:30 - 16:40
A Study on Interior Design Simulation with Multiple CycleGANs
Tatsuya Hagi, Nobutaka Kuroki, Masahiro Numa (Kobe Univ.)

(86)/PRMU 16:40 - 16:50
Category and Ingredients Estimation From Food Image Robust to Typicality of Used Ingredients
Daiki Yamashita, Kengo Murata, Seiya Ito, Kouzou Ohara (Aoyama Gakuin Univ.)

(87)/PRMU 16:50 - 17:00
Comic Speaker Detection and Estimation using Scene Graph
Yingxuan Li, Kiyoharu Aizawa, Yusuke Matsui (UTokyo)

(88)/PRMU 17:00 - 17:10
Continual learning combining Replay and Parameter Isolation Methods
Ryota Adachi, Toshikazu Wada (Wakayama Univ.)

(89) 17:10 - 17:40


----------------------------------------
Fri, Mar 3 PM (16:10 - 17:55)
----------------------------------------

(90)/IBISML 16:10 - 16:25
Dynamic Structure Estimation from Bandit Feedback
Isao Ishikawa (Ehime Univ.), Motoya Ohnishi (Univ. of Washington), Yuko Kuroki (The Univ. of Tokyo), Masahiro Ikeda (RIKEN)

(91)/IBISML 16:25 - 16:40
Fast Identification of Possible Model Parameter Update for Low-Rank Update of Training Data
Hiroyuki Hanada, Noriaki Hashimoto (RIKEN), Kouichi Taji, Ichiro Takeuchi (Nagoya Univ.)

(92)/IBISML 16:40 - 16:55
On the analysis of diverse solutions in topic models
Toshio Uchiyama, Tsukasa Hokimoto (HIU)

(93)/IBISML 16:55 - 17:10
Upper bound of real log canonical threshold based on linear programming problem for the multi-indexes of a polynomial
Joe Hirose (Tokyo Tech)

(94)/IBISML 17:10 - 17:25
Forest Construction of Gaussian and Discrete Variables based on WBIC
Ashraful Islam, Joe Suzuki (Osaka Univ.)

(95)/IBISML 17:25 - 17:40
The free energy of the mean field approximation in normal crossing functions
Miki Aoyagi (Nihon Univ.)

----- Break ( 5 min. ) -----

(96) 17:45 - 17:55


# Information for speakers
General Talk (PRMU/CVIM) will have 10 minutes for presentation.
General Talk (IBISML) will have 10 minutes for presentation and 5 minutes for discussion.


=== Technical Committee on Pattern Recognition and Media Understanding (PRMU) ===
# FUTURE SCHEDULE:

Thu, May 18, 2023 - Fri, May 19, 2023: [Mon, Mar 13]

# SECRETARY:
Contact address: pr-orzer

=== Technical Committee on Information-Based Induction Sciences and Machine Learning (IBISML) ===
# FUTURE SCHEDULE:

Thu, Jun 29, 2023 - Sat, Jul 1, 2023: OIST Conference Center [Wed, May 10]

# SECRETARY:
河原吉伸 ist-u

=== Special Interest Group on Computer Vision and Image Media (IPSJ-CVIM) ===
# FUTURE SCHEDULE:

Thu, May 18, 2023 - Fri, May 19, 2023: [Mon, Mar 13]


Last modified: 2023-03-01 09:44:22


Notification: Mail addresses are partially hidden against SPAM.

[Download Paper's Information (in Japanese)] <-- Press download button after click here.
 
[Cover and Index of IEICE Technical Report by Issue]
 

[Presentation and Participation FAQ] (in Japanese)
 

[Return to PRMU Schedule Page]   /   [Return to IBISML Schedule Page]   /   [Return to IPSJ-CVIM Schedule Page]   /  
 
 Go Top  Go Back   / [HTML] / [HTML(simple)] / [TEXT]  [Japanese] / [English] 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan