Monthly Archives: October 2014

  • Permalink Gallery

    Professor Kuo’s Keynote Speech at LEAP 3.0 Annual Conference on Big Data

Professor Kuo’s Keynote Speech at LEAP 3.0 Annual Conference on Big Data

MCL Director, Professor C.-C. Jay Kuo, gave a keynote speech at the LEAP 3.0 annual conference at the USC Davidson Conference Center at 1:50-2:50pm, October 25 (Saturday), 2014. LEAP is the acronym of “Leadership, Excellence, Aspiration and Platform”, which is a career development forum series launched in 2013. It is co-founded by the Chinese-American Engineers and Scientists Association of Southern California (CESASC) and Southwestern Chinese Students and Scholars Association (SWCSSA). The purpose of the LEAP conferences is to provide a professional networking platform for Chinese students and Chinese American professionals throughout their overseas studies and professional careers.

In his speech, Professor Kuo first gave an overview on big data science and engineering. He pointed out two big data industrial sectors: 1) big data infrastructure providers and 2) big data service providers. Although there will be a steady growth in the sector of big data infrastructure providers, he emphasized the value and opportunities in the sector of big data service providers. Furthermore, he elaborated on several key technologies in big data analytics, including machine learning, data mining, speech understanding and computer vision. At the end of his talk, Professor Kuo made a prediction by saying that “the coming decade (2015-2025) would be the golden decade for the computer vision researchers since quite a few major breakthroughs would come out and lay the foundation of modern computer vision.” Professor Kuo also answered several questions from the floor. The keynote was well received.

By |October 26th, 2014|News|Comments Off on Professor Kuo’s Keynote Speech at LEAP 3.0 Annual Conference on Big Data|

Interview with visiting scholar Dr. Zhengning Wang

In October 2014, MCLab has a new visiting scholar, Dr. Zhengning Wang. Dr. Wang is Associate Professor at School of Electronic Engineering, University of Electronic Science and Technology of China (UESTC), Chengdu, China. We are glad to have an interview with him, talking about his previous research experiences and future expectation at USC.

 

Could you please give an introduction about yourself and your previous research?

I received my Ph.D degree from Southwest Jiaotong University, Chengdu, China in 2007. In 2008, I joined University of Electronic Science and Technology of China and there I worked with about five graduate students. From 2009 to 2011, I worked as a post-doctoral fellow in the second research institute of Civil Aviation Administration of China (CAAC), where I served as a project leader on remote control tower. My research focus is image/video processing, especially video coding, multimedia communication and image enhancement. I also have research experiences in ground surveillance and guidance in air traffic control, video codec design, target tracking for security, light field data compression and high performance computing based on GPU.

 

What is your impression about MCLab and USC?

MCLab has a good atmosphere and I was impressed by the hardworking of the group members. They are also quite friendly and warm-hearted. I received lots of help from Xiaqing, Jian and other group members. USC is a well-known university for her top-notch academic achievements. Researchers here collaborate with each other and exchange their ideas actively, making a perfect opportunity to gain academic communication experiences.

 

What is your expectation and plan in the following academic year?

I hope to work closely with Prof. Kuo to facilitate our research projects. Besides research, I plan to learn about course organization and teaching from USC. I wish [...]

By |October 19th, 2014|News|Comments Off on Interview with visiting scholar Dr. Zhengning Wang|
  • Permalink Gallery

    Professor Kuo’s Keynote Speech at the Optoelectronic Imaging and Multimedia Technology Conference

Professor Kuo’s Keynote Speech at the Optoelectronic Imaging and Multimedia Technology Conference

MCL Director, Professor C.-C. Jay Kuo, gave a keynote speech at the Optoelectronic Imaging and Multimedia Technology Conference, part of 2014 Photonics Asia Program, on October 10, 2014, in Beijing International Convention Center, Beijing, China.

The title of Professor Kuo’s speech was “Big visual data analysis: challenges and solution strategies”. Due to the huge size and great diversity of visual data, big visual data analytics plays a critical role in applications such as large-scale image/video indexing, search and tagging. In this work, Professor Kuo addressed two competing factors that have a high impact on the performance of a classification system; namely, data diversity and data abundance. He presented two strategies to handle the data diversity problem – data grouping and decision stacking. Once the data diversity problem is well resolved, the solution will benefit from data abundance since more data samples allow a learning-based classifier to offer better performance. Professor Kuo used an indoor/outdoor scene classification problem to show the power of the proposed techniques in handling big visual data.

The Optoelectronic Imaging and Multimedia Technology Conference was chaired by Professor Qionghai Dai of Tsinghua University while Photonics Asia was co-organized by the International Society for Optics and Photonics (SPIE) and the Chinese Optical Society (COS). It was the largest annual event showcasing photonics and optical technologies and applications in Asia.

By |October 12th, 2014|News|Comments Off on Professor Kuo’s Keynote Speech at the Optoelectronic Imaging and Multimedia Technology Conference|
  • Permalink Gallery

    Large-scale indoor/outdoor image classification via expert decision fusion

Large-scale indoor/outdoor image classification via expert decision fusion

Scene understanding has been a hot topic in computer vision research fields for quite a long time. As a sub-branch of scene understanding, indoor/outdoor scene classification has become increasingly challenging throughout the years as visual data becomes larger and more diverse. As existing approaches are not scalable to large-scale visual data (usually 100,000 images at least), it is extremely important to have an accurate and robust system to handle big visual data for robust indoor/outdoor classification applications.
Mr. Chen Chen and Ms. Yuzhuo Ren, two MCL PhD students, and Professor C.-C. Jay Kuo recently proposed a new framework to tackle the large-scale indoor/outdoor image classification problem via an expert decision fusion system. The proposed solution has been tested on a large-scale and the most challenging scene understanding dataset from the SUN database, and a correct classification rate of 91.2% has been achieved, which is the state-of-the-art technique in the field. This work has been accepted for publication in the 1st Scene Understanding for Autonomous Systems Workshop (SUAS 2014) held in conjunction with ACCV 2014, Singapore, November 2, 2014.

By |October 5th, 2014|News|Comments Off on Large-scale indoor/outdoor image classification via expert decision fusion|