Thursday, May 29, 2008

Compact Composite Descriptors (CCDs)

Friday, May 23, 2008

Find duplicate images with Dup Detector

The crowd has spoken. We featured an application earlier today that lets you compare images in your library to find dupicate pictures. And a whole bunch of you wrote in to let us know that Dup Detector does the same thing, only better. So we checked it out and you're kind of right, and you're kind of not.
Dup Detector is certainly faster than DupliFinder. And it has many more features. But it's possible it has a few too many features. It took us a couple of minutes to figure out how to load a folder. Once we did, Dup Detector scanned over 2000 images in a bout a minute. When we clicked the Find dups button, the program processed them in a matter of seconds.
For the record, here's how to load a single folder:
Select the Get data tag
Click Build
Click Browse
Navigate to the folder of your choice
Click OK
Selet the Find dups tab
Click Find dups

Select the View Dups tab to actually see the duplicates
You can change the algorithm for detecing similar images. But this is where DupliFinder excels. It does an excellent job of finding images that actually bear a resemblance to one another, while Dup Detector wil spit out a bunch of pictures with similar colors. In other words, if you have a lot of pictures with white backgrounds, Dup Detector will count them as duplicates. You can avoid this by setting the "Dup if within" percentage to something near 100%.

Monday, May 19, 2008

Multimedia Semantics - The Role of Metadata

The first Springer book of the Multimedia Metadata Community is available online. Congratulations to Michael Granitzer, Mathias Lux and Marc Spaniol who did a fantastic job to edit the book in this incredible short period. The book will be in print on February 27, 2008.

Research in the field of multimedia metadata is especially challenging: Lots of scientific publications and reports on research projects are published every year and the range of possible applications is diverse and huge. This book gives an overview on fundamental issues within the field of multimedia metadata focusing on contextualized, ubiquitous, accessible and interoperable services on a higher semantic level. The book in hand provides a selection of basic articles being a base for multimedia metadata research. Furthermore it presents a view on the current state of the art in multimedia metadata research. It provides information from versatile applications domains (Broadcasting, Interactive TV, E-Learning and Social Software) such as:

Multimedia on the Web 2.0 Databases for Multimedia (Meta-)Data Multimedia Information Retrieval & Evaluation
Multimedia Metadata Standards Ontologies for Multimedia
The multimedia metadata community (, where from this book originated, brings together experts from research and industry in the area of multimedia metadata research and application development. The community bridges the gap between an academic research and an industrial scale development of innovative products. By summarizing the work of the community this book contributes to the aforementioned fields by addressing these topics for a broad range of readers.
by Martin Frericks

MPEG-7 & MPEG-21 Community Portal

The MPEG-7 & MPEG-21 community aims at bringing together experts from research and industry in the area of multimedia meta data interoperability for collaborative working environments. By establishing a community of professionals it is intended to bridge the gap between an academic research and an industrial scale development of innovative products for natural collaboration. To accomplish this goal the portal provides information about MPEG-7 & MPEG-21 related events, participating research groups, as well as industry partners to share expertise on multimedia meta data interoperability.

Sunday, May 11, 2008

Moments from WIAMIS 2008

Keynote of H. Bischof

Recently, there has been considerable amount of research in methods for person detecti on. This talk will focus on methods for person detecti on in a surveillance setting (known environment). We will demonstrate that in this setti ng one can build robust and highly reliable person detectors by using on-line learning methods. In parti cular, I will fi rst discuss, conservative learning“ which is able to learn a person detector without any hand labelling effort. As a second example I will discuss a recently developed grid based person detector. The basic idea is to considerably simplify the detecti on problem by considering individual image locations separately. Therefore, we can use simple adapti ve classifi ers which are trained on-line. Due to the reduced complexity we can use a simple update strategy that requires only a few positive samples and is stable by design. This is an essential property for real world applications which require operati on for 24 hours a day, 7 days a week. During the talk we will illustrate our results on video sequences and standard benchmark databases.

Keynote of John R. Smith: Unleashing Video Search

Video is rapidly becoming a regular part of our digital lives. However, its tremendous growth is increasing users’ expectati ons that video will be as easy to search as text. Unfortunately, users are sti ll fi nding it diffi cult to fi nd relevant content. And today’s soluti ons are not keeping pace on problems ranging from video search to content classifi cati on to automati c fi ltering. In this talk we describe recent techniques that leverage the computer’s ability to eff ecti vely analyze visual features of video and apply stati sti cal machine learning techniques to classify video scenes automati cally. We examine related eff orts on the modeling of large video semantic spaces and review public evaluati ons such as TRECVID, which are greatly facilitati ng research and development on video retrieval. We discuss the role of MPEG-7 as a way to store metadata generated for video in a fully standards-based searchable representati on. Overall, we show how these approaches together go a long way to truly unleash video search. (Book af Abstracts)

Thursday, May 8, 2008

New image retrieval engine

The iSemantics image search system is an academic demonstration project trying to show that Internet image search can be made more efficient compared to conventional Internet image search systems.

The purpose of the iSemantics project is to collect semantic links between images and thereby increase the quality of image search results. The image sorting algorithms are taken from the ImageSorter software, developed by Prof. Dr. Kai Uwe Barthel.

Paper is available at WIAMIS 2008 proceedings.