This paper introduces the MSR-Bing grand challenge on image retrieval. The challenge is based on a dataset generated from click logs of a real image search engine. The challenge is to mine semantic knowledge from the dataset and predict the relevance score of any image-query pair. A brief introduction to the dataset, the challenge task, and the evaluation method will be presented. And then the methods proposed by the challenge participants are introduced, followed by evaluation results and some discussions about the goal and future of the challenge.

The semantic gap between low-level visual features and high-level semantics has been investigated for decades but still remains a big challenge in multimedia. When "search" became one of the most frequently used applications, "intent gap", the gap between query expressions and users' search intents, emerged. Researchers have been focusing on three approaches to bridge the semantic and intent gaps: 1) developing more representative features, 2) exploiting better learning approaches or statistical models to represent the semantics, and 3) collecting more training data with better quality. However, it remains a challenge to close the gaps. In this paper, we argue that the massive amount of click data from commercial search engines provides a data set that is unique in the bridging of the semantic and intent gap. Search engines generate millions of click data (a.k.a. image-query pairs), which provide almost "unlimited" yet strong connections between semantics and images, as well as connections between users' intents and queries. To study the intrinsic properties of click data and to investigate how to effectively leverage this huge amount of data to bridge semantic and intent gap is a promising direction to advance multimedia research. In the past, the primary obstacle is that there is no such dataset available to the public research community. This changes as Microsoft has released a new large-scale real-world image click data to public. This paper presents preliminary studies on the power of large-scale click data with a variety of experiments, such as building large-scale concept detectors, tag processing, search, definitive tag detection, intent analysis, etc., with the goal to inspire deeper researches based on this dataset.

Developing an effective vaccine against HIV infection remains an urgent goal. We used a DNA prime/fowlpox virus boost regimen to immunize Chinese rhesus macaques. The animals were challenged intramuscularly with pathogenic molecularly cloned SHIV-KB9. Immunogenicity and protective efficacy of vaccines were investigated by measuring IFN-γ levels, monitoring HIV-specific binding antibodies, examining viral load, and analyzing CD4/CD8 ratio. Results show that, upon challenge, the vaccine group can induce a strong immune response in the body, represented by increased expression of IFN-γ, slow and steady elevated antibody production, reduced peak value of acute viral load, and increase in the average CD4/CD8 ratio. The current research suggests that rapid reaction speed, appropriate response strength, and long-lasting immune response time may be key protection factors for AIDS vaccine. The present study contributes significantly to AIDS vaccine and preclinical research.

Based on the Pichia pastoris secretory expression vector pPIC9K,a recombinant plasmids,pPIC9K-P1/2A-CTB-TEpi,containing capsid polypeptide P1-2A and multi-epitope of FMDV serotype O,have been constructed.Then the expression plasmid was transformed into P.pastoris GS115 by electroporation after lineared with Sac Ⅰ and integrated stably into chromosome of GS115.High-copied transformants GS115/P1/2A-CTB-TEpi was obtained by screen of G418 concentration gradient.Interest protein have been expressed successfully after induced by methanol,and clearaged two fragments,P1/2A and CTB-TEpi,as the autoclasia fragment 2A of FMDV.Molecular weight of the two fragments were 81 800 and 39 400,which accounted for about 25% of the total supernatant protein.The result of Western blotting showed that the two parts of the expression protein could be specifically recognited by serum against FMDV serotype O,and the fragment which contained CTB could be specifically recognited by serum against CT.This work provided a foundation for further study on the evaluation of the multi-epitope subunit vaccine of FMDV serotype O.

This paper presents a context aware, online immediate-mode diagramming recognition and beautification software for hand-sketched diagrams. The system is independent of stroke-order, -number, -direction and is invariant to scaling, translation and rotation. In our stroke-based recognition model, we propose convexity features along with spatial and temporal proximity features to prune the combinatorial search space of possible stroke configurations to form shapes. This reduces the problem of exponential complexity to polynomial one while reducing the error by 24% compared to temporal proximity based criterion. The strokes are then recognized using geometric polygonal features against a neural-net based classifier for 17 classes. The diagramming system is based on stroke-based classifier combination model where an arbitrator makes context aware decisions using suggestions from shape, connector and writing-drawing experts. We achieved an accuracy of 92.7%, 81.4% and 91.5% on the respective experts for a collection of 700,000 online shapes.

objective To develop a multiplex RT-PCR method for determination of foot-and-mouth disease virus(FMDV)type Asia 1 of XJ/03 and JS/05 topotypes.Methods The sequences of a large quantity of epidemic FMDV strains reported in GenBank were compared,based on which the primers for differential diagnosis of infection with FMDV type Asia 1 of XJ/03 and JS/05 topotypes by multiplex RT-PCR were designed,and the conditions for single and multiplex RT-PCR were optimized.The developed method was verified for specificity and used for determination of relevant viruses.Results Both the optimized single and multiplex RT-PCR methods showed high specificities.The determination results of 9 FMDV samples by the developed method were completely consistent with those by gene sequencing.Conclusion A multiplex RT-PCR method for determination of FMDV type Asia 1 of XJ/03 and JS/05 topotypes was developed.which laid a foundation of diagnosis and epidemiological investigation of FMDV as well as application of FMD vaccine.

This paper proposes a machine learning approach to grouping problems in ink parsing. Starting from an initial segmentation, hypotheses are generated by perturbing lo- cal configurations and processed in a high-confidence-first fashion, where the confidence of each hypothesis is pro- duced by a data-driven AdaBoost decision-tree classifier with a set of intuitive features. This framework has success- fully applied to grouping text lines and regions in complex freeform digital ink notes from real TabletPC users. It holds great potential in solving many other grouping problems in the ink parsing and document image analysis domains.

Handwritten text lines are prominent structures in freeform digital ink notes and their reliable detection is the foundation to a natural and intelligent interface for note editing and repurposing. This paper presents an optimization method for text line grouping. The global'cost function is designed to find the simplest stroke partitioning to maximize the likelihood of the resulting lines and the consistency of their configuration. A dynamic programming algorithm provides an initial segmentation of the time-ordered stroke sequence. Then a local gradient-descent algorithm iteratively evaluates splitting and merging hypotheses to minimize the global cost function. On average, the proposed technique processes each note page in less than a second at 90% accuracy.

Handwritten notes are complex structures, which include blocks of text, drawings, and annotations. The main challenge for the newly emerging tablet computer is to provide high-level tools for editing and authoring handwritten documents using a natural interface. One frequent component of natural notes are lists and hierarchical outlines, which correspond directly to the bulleted lists and itemized structures in conventional text, editing tools. We present a system, which automatically recognizes lists and hierarchical outlines in handwritten notes, and then computes the correct structure. This inferred structure provides the foundation for new user interfaces and facilitates the importation of handwritten notes into conventional editing tools.

In order to perform localization and navigation over significant distances (up to a few kilometers), it is important to be able to accurately map the terrain to be traversed. Local methods, such as conventional stereo, do not scale well to large distances and prevent long-range planning. In this paper, we discuss wide-baseline stereo techniques for rovers. In wide-baseline stereo, the image pair is captured with the same camera, but at different positions of the rover. While the larger baseline allows improved accuracy for more distant terrain, stereo matching is more difficult for two reasons. First, odometry errors result in uncertain knowledge of the relative positions of the cameras when the images are captured. Second, the change in perspective makes stereo matching difficult, since image landmarks no longer has the same appearance in both images. We address these problems and show test results on real images.

--This paper presents a new method for estimating piecewise-smooth optical flow. We propose a global optimization formulation with three-frame matching and local variation and develop an efficient technique to minimize the resultant global energy. This technique takes advantage of local gradient, global gradient, and global matching methods and alleviates their limitations. Experiments on various synthetic and real data show that this method achieves highly competitive accuracy.

In this paper we consider the problem of optimal optical flow estimation assuming brightness conservation and piecewise smoothness. We propose a formulation based on three-frame matching and global optimization allowing local variation. It is superior to popular gradient-based models and more justifiable than existing global methods. We also develop an efficient technique to minimize the resultant global energy. It takes advantage of local gradient, global gradient and global matching methods and overcomes their limitations. Experiments on various synthetic and real data and comparison with state-of-the-art techniques show that the method achieves discontinuity preserving capability and sub-pixel accuracy.

In this paper we discuss a hybrid technique for piecewise-smooth optical flow estimation. We first pose optical flow estimation as a gradient-based local regression problem and solve it under a high-breakdown robust criterion. Then taking the output from the first step as the initial guess, we recast the problem in a robust matching-based global optimization framework. We have developed novel fast-converging deterministic algorithms for both optimization problems and incorporated a hierarchical scheme to handle large motions. This technique inherits the good subpixel accuracy from the local gradient approach and the insensitivity to local perturbation and derivative quality from the global matching approach, and it overcomes the limitations of both. Significant advantages over competing techniques are demonstrated on various standard synthetic and real image sequences.

This contest involved the running and evaluation of computer vision and pattern recognition techniques on different data sets with known groundwidth. The contest included three areas; binary shape recognition, symbol recognition and image flow estimation. A package was made available for each area. Each package contained either real images with manual groundtruth or programs to generate data sets of ideal as well as noisy images with known groundtruth. They also contained programs to evaluate the results of an algorithm according to the given groundtruth. These evaluation criteria included the generation of confusion matrices, computation of the misdetection and false alarm rates and other performance measures suitable for the problems. The paper summarizes the data generation for each area and experimental results for a total of six participating algorithms