Computer Vision Laboratory
Graduate school of Systems and Information Engineering,
University of Tsukuba
Welcome to the website of Computer Vision Laboratory!
Computer Vision Laboratory (CVLAB), Graduate school of Systems and Information Engineering, University of Tsukuba
We are always looking for motivated students to help us with our efforts. If you would like to join our lab, please contact with Prof. Fukui via e-mail.
Our laboratory serves as one of the machine learning units in the Center for Artificial Intelligence Research (C-AIR)
Our laboratory also is involved with the Empowerment Informatics Program for Leading Graduate Schools. If you are interested in this program, please feel free to contact Prof. Fukui or Prof. Iizuka with any questions.
Fundamentals and applications of computer graphics. See the homepage of Prof. IIzuka.
Human sensing based on subspace based method, such as face and hand gesture / finger spelling recognition
Subspace based methods: we have been extensively studied the theory and application of subspace analysis for two decades, and developed various types of subspace based method as shown in the left figure. A typical approach in subspace analysis is the subspace method (SM) that classify an input pattern vector into several classes based on the minimum distance or angle between the input pattern vector and each class subspace, where a class subspace corresponds to the distribution of pattern vectors of the class in high dimensional vector space. SM was developed by two Japanese researchers, Watanabe and Iijima around 1970, independently. Watanabe and Iijima named their methods the CLAFIC and the multiple similarity method, respectively. Furthermore, SM has been extended to mutual subspace method (MSM) by replacing an input vector with an input subspace, where the similarity between the input and reference subspace is measured by the canonical angles between them. MSM has well known as one of the most natural and effective classification method for image-sets in pattern recognition and computer vision. We have recently been interested in the theoretical aspect of feature extraction by projecting onto a generalized difference subspace(GDS), which is called GDS projection.
Publications:
2023/03/13 We have a great news. Uchiyama-senpai won the first place on the Open pack challenge:
https://open-pack.github.io/challenge2022/
2023/02/18 Huang's paper titled with "Free-view Expressive Talking Head Video Editing" has been accepted by ICASPP2023. Congratulations!
2022/12/15 Bojan's and our paper titled with "Temporal-stochastic tensor features for action recognition" has been published by the Journal of Machine learning and applications. https://www.sciencedirect.com/science/article/pii/S2666827022000822
2022/10/15 Lincon's and our paper titled with "Grassmannian learning mutual subspace method for image set recognition" has been accepted by the journal of Neurocomputing. Congratulations, Lincon! We are proud of this paper.
2022/08/17 Our paper titled "Visually explaining 3D-CNN predictions for video classification with an adaptive occlusion sensitivity analysis" has been accepted by IEEE WACV2023. Congratulations, Uchiyama-kun! http://arxiv.org/abs/2207.12859
2022/08/16 Our paper titled "Temporal-Stochastic Tensor Features for Action Recognition" has been accepted for the journal of Machine Learning with Applications. Congratulations, Bojan!
2022/07/04 Our paper titled with "Environmental sound classification based on CNN latent subspaces" has been accepted by the 17th International Workshop on Acoustic Signal Enhancement (IWAENC 2022). Congratulations on the acceptance, Maha!
2022/04/12 Our paper titled "Analysis of Temporal Tensor Datasets on Product Grassmann Manifold" has been accepted by the CVPR workshop. Congratulations, Bojan!
2022/04/10 Our paper titled "Discriminant feature extraction by generalized difference subspace" has been accepted for TPAMI. We are proud of this paper.
2022/04/09 Professor Iizuka has received Young research award by the Minister of Education, Culture, Sports, Science and Technology. Congratulations, Iizuka-sensei!
2022/04/08 Our paper titled "SoccerTrack: A Dataset and Tracking Algorithm for Soccer with Fish-eye and Drone Videos" has been accepted by the 8th IEEE International Workshop on Computer Vision in Sports (CVSports). Congratulations, Atom!
2021/12/01 Our paper entitled "Multilinear clustering via tensor Fukunaga–Koontz transform with Fisher eigenspectrum regularization" has been accepted for Applied Soft Computing, Congratulations, Bernardo!
2021/07/05 Our paper entitled ”Constrained mutual convex cone method for image-set based recognition” has been accepted for Pattern recognition.
2021/06/01 Our paper titled "Tensor analysis with n-mode generalized difference subspace" has been accepted for Expert Systems with Applications.
2021/03/12 Our paper entitled "Text classification based on the word subspace representation" has been accepted for Pattern Analysis and Applications.
2020/12/31 Our paper entitled "Tensor analysis with n-mode generalized difference subspace" has been accepted for Experts system and application.
2020/12/01 Our paper entitled "A semi-supervised convolutional neural network based on subspace representation for image classification" has been accepted for EURASIP J. Image Video Process.
2020/07/30 Our paper entitled "Image Harmonization with Attention-based Deep Feature Modulation" has been accepted for BMVC2020.
2020/07/26 Our paper entitled "Discriminative Singular Spectrum Analysis for bioacoustic classification" has been accepted for Interspeech2020.
2020/07/12 Alisher Abdulkhaev (D1) and his teammates from MLT has won a gold medal (3rd) place in the 3rd AI Edge Contest organized by Ministry of Economy, Trade and Industry and NEDO.
2020/05/21 Our paper entitled "Fukunaga-Koontz Convolutional Network with Applications on Character Classification" has been accepted for Neural Process. Letter.
2020/04/16 Our 2 papers entitled "An Interface between Grassmann manifolds and vector spaces" and "Metric Learning with A-based Scalar Product for Image-set Recognition" have been accepted for CVPR 2020 Workshop on Differential Geometry in Computer Vision and Machine Learning.
2020/1/31 We have received Prof. Juan Gabriel Colonna from Federal University of Amazonas as a visiting scholar. We will do an collaborative research on detecting change from time sequential data until the end of Feb.
2020/1/20 We received a visit from three Chinese students of Prof Guijin Wang in Tsinghua University and enjoyed discussion.
2019/10/18 Our paper entitled with "Tensor Fukunaga-Koontz Transform for Hierarchical Clustering" has received the prize for the best paper at the BRACIS 2019.
2019/09/28 Fukuoka-san, who was graduated from our laboratory, has gotten one try in the game with Ireland in the rugby world cup 2019. Congratulations!
2019/09/23 Yoko-san has presented the paper entitled "MLSNET: RESOURCE-EFFICIENT ADAPTIVE INFERENCE WITH MULTI-LEVEL SEGMENTATION NETWORKS" in ICIP2019.
2019/09/20 We have received the visit from Dr. Arturo E. Ceron-Lopez of Hibot Corporation and discussed about the possibility of applications of our approaches to robotics.
2019/09/09 We have received the visit from Dr. Brian Iwata of Kyushu university and have a discussion time on pattern recognition and computer vision.
2019/08/30 Our paper entitled "Enhanced Grassmann Discriminant Analysis with Randomized Time Warping for Motion Recognition " has been accepted for publication in Elsevier's Pattern Recognition.
2019/05/30 Our paper entitled "News2meme: An Automatic Content Generator from News Based on Word Subspaces from Text and Image" has won the best poster award of MVA2019.
2019/05/29 We have presented two papers with the tiles of "News2meme: An Automatic Content Generator from News Based on Word Subspaces from Text and Image" and "Gait Recognition Based on Constrained Mutual Subspace Method with CNN Features" in the sixteenth international conference on machine vision applications, MVA2019.
2019/05/09 We have received the visit from Professor Francois Portet from Laboratoire d'Informatique de Grenoble (LIG) - GETALP - UGA and had got useful advises for our research on text analysis from him.
2019/05/01 Our paper entitled "MLSNET: RESOURCE-EFFICIENT ADAPTIVE INFERENCE WITH MULTI-LEVEL SEGMENTATION NETWORKS" has been accepted for ICIP2019.
2019/04/15 Lincon has presented the paper entitled "Classification of Bioacoustic Signals with Tangent Singular Spectrum Analysis" in ICASSP2019.
2019/04/04 Two international students with the MEXT scholarship have joined the laboratory, coming from Yemen and Serbia.
2019/03/03 Our two papers, which are entitled "News2meme: An Automatic Content Generator from News Based on Word Subspaces from Text and Image" and "Gait Recognition Based on Constrained Mutual Subspace Method with CNN Features", have been accepted for MVA2019.
2019/02/28 Sogi has visited University College London for a month and researched under Dr.Xue's supervision.
Access
Lab Location
University of Tsukuba, The third area, The second floor in E building, 3E205
Address
〒305-8573
University of Tsukuba, Computer Vision Laboratory (Kazuhiro Fukui Lab), 1-1-1 Tennodai,
Tsukuba City, Ibaraki Pref., Japan