Sunday, November 21, 2010

Augsburg dual award adventures

Author: China first victory

ACMMultimedia is most advanced in the field of multimedia session, participants are multimedia content analysis, multimedia applications, systems and interaction of researchers.

Meetings usually include long papers, short papers, presentations and video presentations, and other different plates, which long papers is the most challenging in recent years, every year, only about 50 to 60, the rate of 20 per cent of papers. Microsoft Research Asia since 2003 in this meeting has been very good performance, especially in 2007, the Institute has a total of eight articles long papers accepted, set a new high, representing the total number of papers (57 articles) of 14%. Which I take of the multimedia content analysis and search team has five is received (including University of science and technology cooperation), and I and his Ph.d interns Qi Jun and other four colleagues cooperation after the review of papers was elected to a four-story, one of the best thesis candidates will compete for the best paper award.

To enter the multimedia area international event

2007 September 23, our line of six people sit on the scheduled departure from Munich Airport direct to Augsburg.

This year's multimedia international event will be held here. Shuttle bus from the city, looking out the window, as if to the countryside, farms, ranches and barren land can be seen everywhere. Although there is no point of view, but the air is fresh and translucent and is unparalleled in Beijing. We either chat or simply sleep, enjoy a light and comfortable. But I am uneasy even nervous because I shouldering the expectations of everyone, hope we can win this multimedia areas of top-level International Conference best paper awards.

Selection of the world's top academic conference best paper of candidates is not easy, and best paper of speech is the award-winning thesis is an important factor.

Speak well to let judges convinced your work. As early as one month ago, I began to prepare the report, the work of busy does not allow me to the last few days to prepare. The first walkthrough, but also meticulous preparation of the first version of the slide can be said to be a terrible mess by colleague grant. But it is interesting to note that these researchers for study and research reports critique has an unusual natural reaction, that is, the more the more brave contusion. Criticism of upgrade my passion, my inspiration stimulates criticism. Soon, almost rewritten Second Edition, has been generally recognized. Adoption of the second edition of the proposals, before the final version of the third edition. Interestingly, some students see the slides, it's hard to believe that this is made with PowerPoint – in their words "too funky." I think that in fact "Tuxedo", which is the proper use of animations and effects to aid accurate description of the work. As an intern proposed, one day we may really should write a book about how to use the specialized PowerPoint slides to make an academic report, must be very popular.

In the evening, we arrived at the hotel.

The next morning, as planned, we go at Augsburg College register, and then go to the campus and city acquisition image and video data. In order for the thesis report more attractive and more convincing, we developed a demonstration program in advance, and decide to use local time acquisition of images videos as test data. This is a certain risk, but we would like to try. In the evening, my colleagues helped me organize data, while I was doing last preparations. In fact, the day we work late, in particular, we discovered a bug that caused the demonstration program some data is not good. Some of the FBI for several hours, and finally in the morning before the Council. Good insurance.

In my first report of the General Assembly

"Best paper contest" is the first day of the first technical plates, my report is the first of these, that is, the entire session of the first technical presentations.

Although slightly before, but confident. Half an hour report soon finished, audience members of evaluation is still very good, although the question answer some shortcomings.

Compete for the best essay is a video search and machine learning the latest work, the main research how to use semantic concepts (keywords) linkages to improve the accuracy of the automatic video annotations.

Video annotations are based on the content of the video search key steps, automatic dimensioning good video can use text search technology to index. There are many concept learning methods can be divided into two categories, the concept of independent processing, will be more conceptual learning into multiple classification in two classes. The disadvantage of this type of method is not using semantic concepts. In real-world problems, contact between the semantic concept, and that link can be used to improve the effectiveness of the callout, and you can even take advantage of easy-to-detect semantic concepts to help detect the concept is difficult to detect. The second method is based on the fusion of ways, but the integration is in the first class method based on, or through the integration of separate classification of output results to improve the accuracy of marking. ThisClass methods use a semantic, but this kind of "two-step" policies does not resolve the problem of error propagation, sometimes even reduce annotation results. We propose a concept for both the semantics and the modeling of the relationship between the concept of the new method, called "multiple concepts associated learning" Note 1 (CorrelativeMulti-Label Learning (CML)). This approach to overcome the shortcomings of existing methods, in large datasets on achieved gratifying results.

By Convention, the jury will be discussed at noon, afternoon meeting before the result already had, but have to wait until the next day dinner is announced.

Have tried to ask the knowledge of an insider, resulting not laugh and not for, is to "know" tomorrow. It is precisely because until dinner before I was about to begin, I do not know the following afternoon informed best demo award recommendations to select when you charge the second demo — get the two best prize has no precedent.

Europeans eat really have patience on the second day of the dinner conducted a total of five hours.

The end, we took the best paper award, and also won the best demo award. My boss Li Peng in dinner also does not end in a message sent through networking phone to Beijing.

Unexpected harvest

And best paper Awards come prepared, the best demo award was a surprise, although we also carefully prepared a demo.

The entire conference presentation plate total included 23 shows, apart from Microsoft, IBM, Intel, Yahoo, and other companies and research institutions have also demonstrated their latest research results. Demo showing two Session, covering video image tagging, video browsing, video image retrieval, video ads, mobile devices multimedia applications, and so on. Institute of total wins out of three presentations, which won the "best presentation award" for our group's "video collage".

"Video collage" Note 2 is a compact description of video, it uses an automatic synthesis of images to a summary description of the main content of the video.

For any video, a video collage algorithm to select the best content representation and high visual quality of a video frame, and to take the frame image in the important area, and then apply these regional seamlessly stitching in a specified on the artboard. Relative to the existing video browsing method, the video collage offers a new, more compact, visually more attractive way to browse. Video collage can be easily integrated into existing video management system, video-sharing sites and video search engine to provide more efficient and more pleasant video experience. In particular, the video collage as a video search engine search results summary description, users browse the search results of map, you can quickly understand the contents of the search results to quickly determine whether the user wants to find the video.

This annual meeting of the mainstream papers (approximately one third) is a multimedia search and its related applications, including video and image annotation, video search, video, advertising, media, recommended, and so on.

Research Institute of the eight articles long papers are associated with this topic. This shows that the Internet is booming on the great multimedia areas, as well as academia and industry attaches great importance to media technology in Internet communication, share, search, recommendation and advertising applications. Video surveillance, mobile devices multimedia application, media, interactive, video transmission, etc. are hot topics.

This year also the Institute participated in ACMMultimedia the largest number of years, including twelve interns.

It was also met with several previous research intern, they now are in foreign universities to continue learning, and in this time the General Assembly also has an article published.

Not yet successful, efforts still needed

At the time of this writing, there is good news.

Research Institute has 7 articles on long papers selected for the 2008 ACM Multimedia, which my team still accounts 5, but it's a likely candidate wins best paper. I know, these results more belong to this team, more belongs to the Institute. No Institute such environment, without leaders and boss of the long-term support and guidance, not a group of passionate and energetic team-mate, it is not possible to obtain these. In this area, China academic research level had been questioned. Until now, there are a lot of people, see low-born Chinese researchers. Although our results also limited, to go a long way to go, I and my team-mates will be harder to prove to the world, there is no "gilt" Dr "China" soil "master" of soil can walk in the forefront of world research.

Authors:

China first victory, the network multimedia group research project leader.

He is studying at University for ten years, 2001, Ph.d., the same year joined Microsoft Research Asia, engage in multimedia analysis, search and advertising research. He is a visiting professor at the University of science and technology, or 2008 the MIT Technology Review recognises the world's 35 years of age 35 outstanding young scientists in the world. He often study with students on the road and psychological, learning and life to grow together with the students. He thinks more about philosophy. Often fear the authority, the Authority also often despise. Amateur good calligraphy but can't write, good music and can't sing, good poems without.

Note 1, regarding the specific contents of this method, see: http://research.microsoft.com/~xshua/

Note 2, "video collage" technology, please see: http://research.microsoft.com/~xshua/

No comments:

Post a Comment