Topics

Posted on

Released TanSacNet Pre-release20250105

We updated TanSacNet Project for developing tangent space adaptive control networks.

  • New PyTorch Implementation: Introduced a PyTorch-based 2-D LSUN framework, including batch processing in orthonormalTransform.py and a low-dimensional approximation sample. Optimized gradient computation and sequential matrix processing to improve efficiency in PyTorch and CUDA environments.
  • Enhanced MATLAB Support: Improved data type and device management to ensure stability and performance in MATLAB workflows.
  • Code Refactoring and Stability: Streamlined code structure for maintainability and resolved initialization and CPU processing issues across both frameworks.

This pre-release highlights the new PyTorch implementation, along with key updates to MATLAB support and overall code stability.

Paper Published in IAPSIPA Transactions on Signal and Information Processing

 

Posted on

Paper Published in ITE Transactions on Media Technology and Applications

The following paper has been published in ITE Transactions on Media Technology and Applications.

  • Jikai Li, Shogo Muramatsu, [Paper] Structured Deep Image Prior for Image Denoising with Interscale SURE-LET, ITE Transactions on Media Technology and Applications, 2025, Volume 13, Issue 1, Pages 187-199, Released on J-STAGE January 01, 2025, Online ISSN 2186-7364, https://doi.org/10.3169/mta.13.187, https://www.jstage.jst.go.jp/article/mta/13/1/13_187/_article/-char/en,
  • Abstract:
    This study develops a self-supervised image denoising technique that incorporates a structured deep image prior (DIP) approach with Stein’s unbiased risk estimator and linear expansion of thresholding (SURE-LET). Leveraging interscale and interchannel dependencies of images to develop a multichannel denoising approach. The original DIP, introduced by Ulyanov et al. in 2018, requires a random image as the input for restoration, offering an advantage of not requesting training data. However, the interpretability of the role of the network is limited, and challenges exist in customizing its architecture to incorporate domain knowledge. This work integrates SURE-LET with Monte Carlo computation into the DIP framework, providing the reason of the random image supply and shifting the focus from generator to restorer design, thus enabling the network structure of DIP to more easily reflect domain knowledge. The significance of the developed method is confirmed through denoising simulations using the Kodak image dataset.

 

Posted on

APSIPA ASC 2024 Participation Report

APSIPA ASC 2024 was held at Galaxy International Convention Center, Macau, China, from December 3 to 6, 2024.

We gave the following presentation.

I had a valuable opportunity to deliver an oral presentation on-site.
I also had a highly meaningful experience, gaining valuable insights through interactions with individuals external to the organization.

Next year’s APSIPA ASC 2025 will be held in Singapore!

Posted on

The 7th Symposium on Civil Engineering Science Participation Report

The 7th Symposium on Civil Engineering Science (organized by the ARCE Project) was held at Toki Messe on Friday, November 8, 2024.

The following poster presentations were made from our laboratory:

〇Phonepaserth Sisaykeo (M2)：Study on Digital Twin of River by 3D Modeling for Flow Path Health Assessment

〇Ryuto Ito (M1)：河川水位一体制御のための有向グラフ信号処理

〇Ryusei Aoki (B4)：ROS 2を用いた河川流路制御プロトタイプシステムの構築

〇Hiromu Kanauchi (B4)：河川水位一体制御のためのデータ駆動による水位分布予測手法の性能評価

The 7th Civil Engineering Science Symposium was held under the theme “Thinking about People and Rivers 100 Years from Now.” From the standpoints of industry, government, and academia, the symposium provided an opportunity for in-depth discussions on the ideal state of people and rivers 100 years from now, current constraints, and the respective demands and task setting of industry, government, and academia.

Our laboratory also made a poster presentation on the results of our research from the viewpoint of signal processing, and we were able to exchange valuable opinions with people from various fields of expertise.

We will continue to work hard on our research so that we can realize the ideal of people and rivers as soon as possible.

Posted on

ITE Technical Group Meeting on Media Engineering (ME) Oct. 2024 Participation Report

The Institute of Image Information and Television Engineers (ITE) Technical Group Meeting on Media Engineering (ME) was held on October 1st-2nd, 2024, at Room 1, B3F, Japan Society for the Promotion of Machine Industry (3-5-8 Shiba-Koen, Minato-ku, Tokyo 105-0011, Japan).

We gave the following presentations.

Through this presentation, we engaged in meaningful discussions with professors and specialists in media engineering and digital modeling, gaining insights into the latest advancements in Digital Twin applications. We were excited to receive valuable feedback and learn that our research is drawing interest from researchers across various fields and institutions.

It was an enriching experience to participate in presentations that addressed diverse approaches in media engineering, offering us fresh perspectives and inspiration for future research.

 

Posted on

2024 IEEE Shin-etsu Conference Session, Niigata City, September 28, 2024.

2024 IEEE Shin-etsu Branch Conference Session was held on Saturday, September 28, 2024, at Niigata University, Niigata City, Japan).

We Gave the Following Oral Presentation.

  • Tukur SADA1，Shogo MURAMATSU2， (Niigata University) : Graph Signal Processing Application for Enhancing Reliability and Stability of Electrical Power Grid.

Similarly, at 2024 Shin-etsu Chapter Conference, Social gathering and student award ceremony was held on Saturday evening, September 28, 2024, at Niigata University.

Posted on

Niigata University Data Science and AI Symposium 2024 Participation Report

The Niigata University Data Science and AI Symposium 2024 was held on Thursday, September 19, 2024, at Bandai Silver Hotel (1-3-30 Bandai, Chuo-ku, Niigata City, Japan).

We gave the following poster presentation.

    • Tsubasa NAITO1， Ryuto ITO1， Yuichi TANAKA2， Shogo MURAMATSU1 (1. Niigata Univ.，2. Osaka Univ.) : DICTIONARY LEARNING FOR DIRECTED GRAPH SIGNALS VIA AUGMENTED GFT

    • Phonepaserth Sisaykeo1, Hiroyasu Yasuda1, Kiyoshi Hayasaka1 and Shogo Muramatsu1 (1. Niigata University) :  FLOW-PATH FITTING FROM IMAGES WITH SUPPORT VECTOR REGRESSION FOR RIVER HEALTH ASSESSMENT

    • Ryuto Ito1, Tsubasa Naito1, Hiroyasu Yasuda1, Masaaki Nagahara2 and Shogo Muramatsu1 (1. Niigata Univ., 2. Hiroshima Univ.) :  SPARSE-CODED TIME-DELAY GRAPH DMD FOR PREDICTION OF RIVER WATER LEVEL DISTRIBUTION

We engaged in insightful discussions on the latest trends in Generative AI, Big Data, and cloud technologies. The presentations sparked valuable exchanges, and the activity report provided a comprehensive view of cutting-edge research applications. The poster session also offered an opportunity to receive useful feedback through a lively exchange of ideas with the participants.

Overall, it was a stimulating experience, highlighting the significance of our work and fostering opportunities for future collaboration.