Saturday, November 29, 2008
The first VideoTagGame ran back in summer of 2007 during a Yahoo! party in Amsterdam. Now they're ready to take their experiment to the public through the Yahoo! Sandbox so they can collect more statistics on its usage.
The objective of the VideoTagGame is to collect time-based annotations of the video which could then enable the retrieval of relevant parts in a video when a search is performed, rather than returning the entire video itself. These annotations are collected in the context of a multi-player game.
How To Play
To play the VideoTagGame, participants must sign in with their Yahoo! ID and join a new game. There will always be at least three players in each game. After a 3-second countdown, the video will begin to play. As it plays, participants enter tags that correspond to the various parts of the video. When two players agree on a tag (that is, they enter the same tag), they each get points. The closer together the tags were entered, the more points are rewarded. After the video ends, participants can then watch as it plays again, this time with the tags overlaid on top of the video.
Friday, November 28, 2008
Yottalook™ is based on core technologies developed by iVirtuoso to achieve optimized search results. First is automated analysis of the search term to understand what the radiolgist is trying to look for - this core technology is called "natural query analysis".
Yottalook™ has also developed a thesaurus of medical terminologies that not only identifies synonyms of terms but also defines relationships between terms. This second core technology is called "semantic ontology" and is based on existing medical ontologies that have been enhanced by iVirtuoso, such as RadLex - A Lexicon for Uniform Indexing and Retrieval of Radiology Information Resources developed by the Radiology Society of North America.
Third core technology is "relevance algorithm" for image search that differentiates medical terms from other words in text associated with medical images and uses them to create ranking for Yottalook image search.
The fourth core technology is a specialized content delivery system called "Yottalinks" that provides high yield content based on the search term. This content may also be provided by a third party vendor licensing Yottalook search. Yottalook™ can be integrated with any web based medical application so that context relevant information is provided to the physician at the point of care. http://www.yottalook.com/
Thursday, November 27, 2008
Tuesday, November 25, 2008
The new version features better visualization techniques for all the kinds of screens, plus, a new feature that preserves even better the correctly exposed image regions. PhotoEnhancer 2.2 is the most complete version of this project.
Monday, November 24, 2008
Jena is open source and grown out of work with the HP Labs Semantic Web Programme.
The Jena Framework includes:
A RDF API
Reading and writing RDF in RDF/XML, N3 and N-Triples
An OWL API
In-memory and persistent storage
SPARQL query engine
Sunday, November 23, 2008
CVPR 2009 will be held at the Fontainebleau Hotel in Miami, Florida.
Papers in the main technical program must describe high-quality, original research.
Topics of interest include all aspects of computer vision and pattern recognition (applied to images and video) including, but not limited to, the following areas:
Early and Biologically-inspired Vision
Color and Texture
Segmentation and Grouping
Computational Photography and Video
Motion and Tracking
Stereo and Structure from Motion
Illumination and Reflectance Modeling
Shape Representation and Matching
Object Detection, Recognition, and Categorization
Video Analysis and Event Recognition
Face and Gesture Analysis
Statistical Methods and Learning
Medical Image Analysis
Image and Video Retrieval
Vision for Graphics
Vision for Robotics
Vision for Internet
Applications of Computer Vision
Saturday, November 22, 2008
Topics of interest include, but are not limited to:
Multimedia indexing and retrieval (image, audio, video, text)
Matching and similarity search
Construction of high level indices
Multimedia content extraction
Identification and tracking of semantic regions in scenes
Multi-modal and cross-modal indexing
Multimedia data mining
Metadata generation, coding and transformation
Large scale multimedia database management
Summarisation, browsing and organization of multimedia content
Presentation and visualization tools
User interaction and relevance feedback
Personalization and content adaptation
Evaluation and metrics
Thursday, November 20, 2008
July 8-10, 2009, Santorini Island, Greece - http://www.civr2009.org/ -
Image and Video retrieval have now reached a state where successful techniques and applications start flourishing. The ACM International Conference on Image and Video Retrieval (ACM-CIVR) series of conferences is the ideal opportunity to present and encounter such developments. Originally set up to illuminate the state-of-the-art in image and video retrieval throughout the world, it is now a reference event in the field where researchers and practitioner exchange knowledge and ideas. CIVR2009 is seeking original high quality special sessions addressing innovative research in the broad field of image and video retrieval. We wish to highlight significant and emerging areas of the main problem of search and retrieval but also the equally important related issues of multimedia content management, user interaction and community-based management.
Example topics of interest include but are not limited to: social network information mining, unsupervised methods for data exploration, large scale issues for algorithms and data set generation.
Each special session will consist of 5 invited papers. The organizers’ role is to attract the speakers and chair the session itself. Proposals will be evaluated based on the timeliness of the topic, relevance to CIVR, the degrees to which they will bring together key researchers in the area, introduce the area to the larger research community, further develop the area, and potentials to establish a larger community around the area. Please note that all papers in the proposed session will undergo the same review process as regular papers. If after the reviewing process less than the necessary number of papers solicited for a special session are selected, the Special Session will be cancelled, and the solicited papers that passed review process will be presented within regular sessions of the conference.
Microsoft Live Labs has turned these research ideas into a streaming multi-resolution Web-based service called Photosynth.
You can also read about newer research we have been doing in this area at the University of Washington Photo Tourism project page.
Paper and video
Noah Snavely, Steven M. Seitz, and Richard Szeliski, Photo Tourism: Exploring photo collections in 3D," ACM Transactions on Graphics, 25(3), August 2006. (Video WMV), Video (MOV))Abstract
We have developed a system for interactively browsing and exploring large unstructured collections of photographs of a scene using a novel 3D interface. Our system consists of an image-based modeling front end, which automatically computes the viewpoint of each photograph as well as a sparse 3D model of the scene and image to model correspondences. Our photo navigation tool uses image-based rendering techniques to smoothly transition between photographs, while also enabling full 3D navigation and exploration of the set of images and world geometry, along with auxiliary information such as overhead maps. Our system also makes it easy to construct photo tours of scenic or historic locations, as well as to annotate image details, which are automatically transferred to other relevant images in the collection. We demonstrate our system on several large personal photo collections as well as images gathered from photo sharing Web sites on the Internet.
Wednesday, November 19, 2008
"Our goal is to develop an easy-to-use computaion and simulation platform with a C++ like syntax. We have adopted Matlab's structure philoshophy and C++ 's structured language syntax. There are various toolboxes (packages of functions relative to a certain scientific field), which depend on open-source libraries." EngLab Team
The EngLab distribution is available in two ways: there are two basic Englab releases, EngLab Console and EngLab GUI. EngLab Console allows EngLab's execution through the console(Linux or Windows). EngLab GUI gives the opportunity of using EngLab through a graphical user interface. EngLab GUI is implemented with the use of the open-source library wxWidgets 2.8, providing additional usability compared to EngLab Console edition. EngLab GUI is independent, so there is no need for EngLab Console to be installed, in order to properly install and execute EngLab GUI.
Toolboxes are distributed as seperate packages. Their installation is possible either through EngLab Console or EngLab GUI. The reason is that those toolboxes depend on open-source libraries that have to be previously installed. So as the user not to be forced to install those libraries directly, user can install packages and toolboxes at his/her own will.
For the time being, EngLab Console edition is available for Windows and Linux and Englab GUI is available for Linux only.
Until now EngLab has the following features :
- 16 types of variable declaration (int, float, ...)
- Variable declaration with unlimited number of dimensions.
- Loop structures (for, while, ...)
- Arithmetic, logical and binary operations
- Constant number declaration (pi, phi, ...)
- Graphical manipulation of variable values of any dimension (Englab GUI)
- Adjustable graphical environment (Englab GUI)
- Editor for writing *.eng functions (Englab GUI)
- Command history for the last 5 sessions
- Immediate access to variables, constants and functions (EngLab GUI)
- Recent files opened through EngLab (EngLab GUI)
Toolboxes that have been fully or partially implemented:
- a package containing fundamental functions of C (trigonemetric, hyperbolic trigonometrical, ...)
- a package containing some statistic functions
- a package containing functions that allow convertions of the variable type
All these toolboxes accompany the basic two EngLab editions, since they do not depend on another open-source library. Moreover, some other toolboxes have been partially implemented:
- a package that contains functions for the manipulation of 2-D matrices (determinant, inverse array, ...). This package depends on the open-source library NewMat10.
- a package that contains functions for image processing. This package depends on the open-source library CImg.
- a package that contains functions for image processing. This package depends on the open-source library OpenCV.
- a toolbox for visual data representation(plots etc)
- a toolbox that contains functins for manipulating polyonymials, root detection, computation of integrals and derivatives, special functions and more.
Tuesday, November 18, 2008
The technology works using 10 stock human emotions - for instance happiness, sadness, concern - that have been programmed into the robot.
The software then maps what it sees to Jules' face to combine expressions instantly to mimic those being shown by a human subject.
Controlled only by its own software, Jules can grin and grimace, furrow its brow, and 'speak' as the software translates real expressions observed through video camera 'eyes'.
If you want people to be able to interact with machines, then you've got to be able to do it naturally...When it moves it has to look natural in the same way that human expressions are, to make interaction useful.Chris Melhuish, head of the Bristol Robotics Laboratory
The robot - made by US roboticist David Hanson - then copies the facial expressions of the human by converting the video image into digital commands that make the robot's inner workings produce mirrored movements.
And it all happens in real time as Jules is bright enough to interpret the commands at 25 frames per second.
The project was developed over more than three years at the Bristol Robotics Laboratory, a lab run by the University of the West of England and the University of Bristol under the leadership of Chris Melhuish, Neill Campbell and Peter Jaeckel.
The aim of the developers was to make it easier for humans to interact with 'artificial intelligence', in other words to create a 'feelgood' factor.
The BRL's Peter Jaeckel said: ''Realistic, life-like robot appearance is crucial for sophisticated face-to-face robot - human interaction.
''Researchers predict that one day robotic companions will work, or assist humans in space, care and education. Robot appearance and behaviour need to be well matched to meet expectations formed by our social experience."
But a warning note has been sounded.
Kerstin Dautenhahn, a robotics researcher at the University of Herefordshire, believes that people may be disconcerted by humanoid automatons that simply look 'too human'.
''People might easily be fooled into thinking that this robot not only looks like a human and behaves like a human, but that it can also feel like a human. And that's not true," she pointed out.
Monday, November 17, 2008
Friday, November 14, 2008
1. Draw a sketch to retrieve similar images from our database. The method is based on a new spatial compact color descriptor.
2. Automatic keyword annotation. Select a combination of words and retrieve images. The method is based on a fuzzy support vector machine system. The network was trained using a combination of CEDD and FCTH descriptors
Thursday, November 13, 2008
Friendships are shown as dark gray lines, and common photo appearances are shown as a lighter gray line with a number in the center. The number indicates how many photos the two people appeared in together.
Personal vs. Friends social networks
* When launched from one's own profile, information about all of one's friends and their friendships is loaded.* When launched from another user's profile (Using the "TouchGraph Friends + Photos" link below their profile picture) only people tagged in their photos will appear in the graph.
One can not see another person's whole social network because Facebook only allows applications to get a list of one's own friends. For other users it is only possible to get a list of people that they appear in photos with. Perhaps Facebook's policy will change in the future.
The TouchGraph Facebook Browser determines the clusters/cliques to which one's Friends belong and uses different colors to show each clique. Cliques are characterized by having lots of friendships within a group of friends and few connections to members outside the group.
Friends are assigned a Rank so that one can reduce clutter by only showing a set of 'Top' friends. TouchGraph gives the highest rank to friends who are connectors between different cliques. Finding connecters involves a metric called Betweeness Centrality which is an established measure for a person's importance within a social network.
Wednesday, November 12, 2008
Sunday, November 9, 2008
PhotoEnhancer is an experimental image enhancement software, which employs the characteristics of the ganglion cells of the Human Visual System. Many times the image captured by a camera and the image in our eyes are dramatically different. Especially when there are shadows or highlights in the same scene. In these cases our eyes can distinguish many details in the shadows or the highlights, while the image captured by the camera suffers from loss of visual information in these regions.
PhotoEnhancer attempts to bridge the gap between "what you see" and "what the camera sees". It enhances the shadow or the highlight regions of an image, while keeping intact all the correctly exposed ones. The final result is a lot closer to the human perception of the scene, than the original captured image, revealing visual information that otherwise wouldn't be available to the human observer.
The latest version of PhotoEnhancer (2.4) has been released. Version 2.4 features a 'Batch Processing' mode, for the quick enhancement of many image files just with a few clicks, as well as an improved user interface.
The new version of PhotoEnhancer features the method of "Multi-Scale Image Contrast Enhancement". This additional algorithm enhances locally the contrast of images, maximizing the available visual information. It can be applied to foggy scenes, aerial or satellite images, images with smoke or medical images.
Bugs report and suggestions at: firstname.lastname@example.org
Saturday, November 8, 2008
Wednesday, November 5, 2008
X. Anguera, N. Oliver, and M. Cherubini. Multimodal and mobile personal image retrieval: A user study. In K. L. Chan, editor, Prooceeding of the International Workshop on Mobile Information Retrieval (MobIR’08), pages 17–23, Singapore, 20-24 July 2008. [PDF]
Mobile phones have become multimedia devices. Therefore it is not uncommon to observe users capturing photos and videos on their mobile phones. As the amount of digital multimedia content expands, it becomes increasingly difficult to find specific images in the device. In this paper, we present our experience with MAMI, a mobile phone prototype that allows users to annotate and search for digital photos on their camera phone via speech input. MAMI is implemented as a mobile application that runs in real-time on the phone. Users can add speech annotations at the time of capturing photos or at a later time. Additional metadata is also stored with the photos, such as location, user identification, date and time of capture and image-based features. Users can search for photos in their personal repository by means of speech without the need of connectivity to a server. In this paper, we focus on our findings from a user study aimed at comparing the efficacy of the search and the ease-of-use and desirability of the MAMI prototype when compared to the standard image browser available on mobile phones today.
In this article, a C# library for neural network computations is described. The library implements several popular neural network architectures and their training algorithms, like Back Propagation, Kohonen Self-Organizing Map, Elastic Network, Delta Rule Learning, and Perceptron Learning. The usage of the library is demonstrated on several samples:
Classification (one-layer neural network trained with perceptron learning algorithms);
Approximation (multi-layer neural network trained with back propagation learning algorithm);
Time Series Prediction (multi-layer neural network trained with back propagation learning algorithm);
Color Clusterization (Kohonen Self-Organizing Map);
Traveling Salesman Problem (Elastic Network).
The attached archives contain source codes for the entire library, all the above listed samples, and some additional samples which are not listed and discussed in the article.
The article is not intended to provide the entire theory of neural networks, which can be found easily on the great range of different resources all over the Internet, and on CodeProject as well. Instead of this, the article assumes that the reader has general knowledge of neural networks, and that is why the aim of the article is to discuss a C# library for neural network computations and its application to different problems.
Intelligent Decision Technologies (IDT) journal seeks original manuscripts
for a Special Issue on Advances in Medical Decision Support Systems scheduled to appear in Vol. 3, No. 2, 2009.The last few decades have witnessed significant advancements in intelligent computation techniques. Driven by the need to solve complex real-world problems, powerful and sophisticated intelligent data analysis technologies have been exploited or emerged, such as neural networks, support vector machines, evolutionary algorithms, clustering methods, fuzzy logic, particle swarm optimization, data mining, etc. In recent years, the volume of biological data has been increasing exponentially, thus, allowing significant learning and experimentation to be carried out using a multidisciplinary approach, which gives rise to many challenging problems. The foundation for any medical decision support is the medical knowledge base which contains the necessary rules and facts. This knowledge needs to be acquired from information and data in the fields of interest, such as medicine. Clinical decision-making is a challenging, multifaceted process. Its goals are precision in diagnosis and institution of efficacious treatment. Achieving these objectives involves access to pertinent data and application of previous knowledge to the analysis of new data in order to recognise patterns and relations. As the volume and complexity of data have increased, use of digital computers to support data analysis has become a necessity. In addition to computerisationof standard statistical analysis, several other techniques for computer-aided data classification and reduction generally referred to as intelligent systems, have evolved. This special issue will focus on illustrative and detailed information about medical intelligent decision support systems and feature extraction/selection for automated diagnostic systems.The focus of this special issue is on advances in medical intelligent decision support systems including determination of optimum classification schemes for the problem under study and also to infer clues about the extracted features. Topics include, but are not limited to, the following:
* Bioinformatics and Computational Biology
* Neural Networks, Fuzzy Logic Systems and Support Vector Machines in Biological Signal Processing
* Decision Support Systems and Computer Aided Diagnosis
* Biomedical Signal Processing
* Biomedical Imaging and Image Processing
* Modelling, Simulation, Systems, and Control
Paper submission: Submitted articles must not have been previously published or currently submitted for journal publication elsewhere. As an author, you are responsible for understanding and adhering to our submission guidelines. You can access them from http://www.iospress.nl. Please thoroughly read these before submitting your manuscript. Each paper will go through a rigorous review process.
Please note the following important dates:
Paper submission for review: November 30, 2008 (Final deadline)
Review results: January 15, 2009
Revised Paper submission: February 20, 2009
Final acceptance: March 1, 2009
Manuscript delivery to the publisher: April 15, 2009
Interested authors should submit digital copies (PDF preferred) of their papers (suggested paper length: 15 pages), including all tables, diagrams, and illustrations, to the Guest Editor, Dr. Vassilis S. Kodogiannis, bye-mail.