Time filter

Source Type

Rochester, NY, United States

Joshi D.,Eastman Kodak Co. | Datta R.,Pennsylvania State University | Fedorovskaya E.,Moscow State University | Wang J.Z.,University of Minnesota | And 2 more authors.
IEEE Signal Processing Magazine | Year: 2011

In this tutorial, we define and discuss key aspects of the problem of computational inference of aesthetics and emotion from images. We begin with a background discussion on philosophy, photography, paintings, visual arts, and psychology. This is followed by introduction of a set of key computational problems that the research community has been striving to solve and the computational framework required for solving them. We also describe data sets available for performing assessment and outline several real-world applications where research in this domain can be employed. © 2011 IEEE. Source

Yang Y.,Zhejiang University | Zhuang Y.,Zhejiang University | Tao D.,Intelligent Systems Technology, Inc. | Xu D.,Nanyang Technological University | And 2 more authors.
IEEE Transactions on Circuits and Systems for Video Technology | Year: 2010

In this paper, we propose a new method to recognize gestures of cartoon images with two practical applications, i.e., content-based cartoon image retrieval and interactive cartoon clip synthesis. Upon analyzing the unique properties of four types of features including global color histogram, local color histogram (LCH), edge feature (EF), and motion direction feature (MDF), we propose to employ different features for different purposes and in various phases. We use EF to define a graph and then refine its local structure by LCH. Based on this graph, we adopt a transductive learning algorithm to construct local patches for each cartoon image. A spectral method is then proposed to optimize the local structure of each patch and then align these patches globally. MDF is fused with EF and LCH and a cartoon gesture space is constructed for cartoon image gesture recognition. We apply the proposed method to content-based cartoon image retrieval and interactive cartoon clip synthesis. The experiments demonstrate the effectiveness of our method. © 2006 IEEE. Source

Chen H.,Stanford University | Gallagher A.,Kodak Research Laboratories | Gallagher A.,Cornell University | Girod B.,Stanford University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2012

Describing clothing appearance with semantic attributes is an appealing technique for many important applications. In this paper, we propose a fully automated system that is capable of generating a list of nameable attributes for clothes on human body in unconstrained images. We extract low-level features in a pose-adaptive manner, and combine complementary features for learning attribute classifiers. Mutual dependencies between the attributes are then explored by a Conditional Random Field to further improve the predictions from independent classifiers. We validate the performance of our system on a challenging clothing attribute dataset, and introduce a novel application of dressing style analysis that utilizes the semantic attributes produced by our system. © 2012 Springer-Verlag. Source

Cao L.,IBM | Jin X.,University of Illinois at Urbana - Champaign | Yin Z.,University of Illinois at Urbana - Champaign | Del Pozo A.,University of Illinois at Urbana - Champaign | And 3 more authors.
Neurocomputing | Year: 2012

Random walk was first introduced by Karl Pearson in 1905 and has inspired many research works in different fields. In recent years, random walk has been adopted in information network research, for example, ranking and similarity estimation. In this paper, we introduce a new model called RankCompete, which allows multiple random walkers in the same network (existing work mostly focus on random walks of a single category). By introducing the "competition" concept into the random walk framework, our method can fulfill both clustering and ranking tasks simultaneously and thus provide an effective analysis tool for networks. Compared with the traditional network ranking approaches, our new method focuses more on the structure of specialized clusters. Compared with the traditional graph clustering approaches, our new method provides a faster and more intuitive way to group the network nodes. We validate our approach on both bibliography networks and visual information networks, and the results show that our approach can obtain 100% perfect clustering results in clustering the DBLP 20 conferences and outperform the state-of-the-art on the Cora dataset. Furthermore, we show that our method can be effectively used to summarize personal photo collections. © 2012 Elsevier B.V.. Source

Milliman H.W.,Case Western Reserve University | Boris D.,Kodak Research Laboratories | Schiraldi D.A.,Case Western Reserve University
Macromolecules | Year: 2012

Polyhedral oligomeric silsesquioxanes (POSS) have been incorporated into a wide range of polymers over the past two decades in an attempt to enhance their thermal and mechanical properties. Properties of POSS/polymer blends/composites are highly dependent on the uniformity of POSS dispersion and thus are particularly sensitive to the magnitude of interaction between POSS and added fillers/polymers. Methods to characterize these interactions in terms of solubility parameters have been recently examined in the literature using group contribution calculations. The present work presents a method for measuring three-dimensional Hansen solubility parameters for polymers and POSS which allows for the direct calculation of interaction potentials. These measured solubility parameters predict POSS/polymer interactions more accurately than group contribution calculations and accurately predict the uniformity of POSS dispersion and the resultant property enhancements. © 2012 American Chemical Society. Source

Discover hidden collaborations