The proceedings contain 525 papers. The topics discussed include: learning visual similarity measures for comparing never seen objects;a contextual dissimilarity measure for accurate and efficient image search;learnin...
详细信息
ISBN:
(纸本)1424411807
The proceedings contain 525 papers. The topics discussed include: learning visual similarity measures for comparing never seen objects;a contextual dissimilarity measure for accurate and efficient image search;learning local image descriptors;principal curvature-based region detector for object recognition;a benchmark for the comparison of 3-D motion segmentation algorithms;a nonparametric treatment for location/segmentation based visual tracking;learning gaussian conditional random fields for low-level vision;hierarchical structuring of data on manifolds;learning GMRF structures for spatial priors;element rearrangement for tensor-based subspace learning;unsupervised clustering using multi-resolution perceptual grouping;object tracking by asymmetric kernel mean shift with automatic scale and orientation selection;and free-form nonrigid image registration using generalized elastic nets.
We propose the creation of an e-research facility with the working title European laboratory for software evolution (ELSE). Our motivation is that a significant obstacle to progress in researching evolutionary phenome...
详细信息
We propose the creation of an e-research facility with the working title European laboratory for software evolution (ELSE). Our motivation is that a significant obstacle to progress in researching evolutionary phenomena in software-intensive systems is the difficulty of conducting realistic, industrial-scale investigations, especially for small research teams operating at armpsilas-length from each other. ELSE will address this problem by providing a virtual research environment (VRE) where researchers in software evolution and related fields can find annotated collections of data, tools, patterns and templates, dasiahow-topsila documents, previous results, and access to expertise in using them. The Laboratory will provide an on-line environment where researchers and practitioners can share research facilities and experiences and seek partners in projects. ELSEpsilas concerns will encompass both the social and the engineering aspects of evolution in software-intensive systems, and will therefore draw on existing achievements in both e-social science and e-science. It will create an exciting and innovative facility that will be attractive to researchers in fields ranging from software engineering, through informatics and business history to society and technology studies and anthropology. We are actively seeking both academic and industrial partners who will join us in taking this vision forward.
In this paper we apply approximate and multiscale entropy metrics to spectrum occupancy data gathered during an extensive measurement campaign. We show that the presented methods can be successfully applied to search ...
详细信息
In this paper we apply approximate and multiscale entropy metrics to spectrum occupancy data gathered during an extensive measurement campaign. We show that the presented methods can be successfully applied to search for and quantify structures of highly varying complexities and time scales. Although structures can be found they are unfortunately relatively complex and it does not appear to be easy to directly exploit them for dynamic spectrum access. We also discuss the major foreseen application areas for entropy-based analysis. These include increasing the reliability of spectrum sensing as well as validation of spectrum occupancy models. Finally, we highlight some results that should be taken into account in future work on spectrum sensing. In particular we observe that strong interference has significant structure, and cannot be well approximated by noise.
Current methods for learning visual categories work well when a large amount of labeled data is available, but can run into severe difficulties when the number of labeled examples is small. When labeled data is scarce...
详细信息
ISBN:
(纸本)9781424411795
Current methods for learning visual categories work well when a large amount of labeled data is available, but can run into severe difficulties when the number of labeled examples is small. When labeled data is scarce it may be beneficial to use unlabeled data to learn an image representation that is low-dimensional, but nevertheless captures the information required to discriminate between image categories. This paper describes a method for learning representations from large quantities of unlabeled images which have associated captions;the goal is to improve learning in future image classification problems. Experiments show that our method significantly outperforms (1) a fully-supervised baseline model, (2) a model that ignores the captions and learns a visual representation by performing PCA on the unlabeled images alone and (3) a model that uses the output of word classifiers trained using captions and unlabeled data. Our current work concentrates on captions as the source of meta-data, but more generally other types of meta-data could be used.
We propose a novel classification approach for automatically detecting pulmonary embolism (PE) from computed-tomography-angiography images. Unlike most existing approaches that require vessel segmentation to restrict ...
详细信息
ISBN:
(纸本)9781424411795
We propose a novel classification approach for automatically detecting pulmonary embolism (PE) from computed-tomography-angiography images. Unlike most existing approaches that require vessel segmentation to restrict the search space for PEs, our toboggan-based candidate generator is capable of searching the entire lung for any suspicious regions quickly and efficiently. We then exploit the spatial information supplied in the vascular structure as a post-candidate-generation step by designing classifiers with geodesic distances between candidates along the vascular tree. Moreover, a PE represents a cluster of voxels in an image, and thus multiple candidates can be associated with a single PE and the PE is identified if any of its candidates is correctly classified. The proposed algorithm also provides an efficient solution to the problem of learning with multiple positive instances. Our clinical studies with 177 clinical cases demonstrate that the proposed approach outperforms existing detection methods, achieving 81% sensitivity on an independent test set at 4 false positives per study.
The proceedings contain 162 papers. The topics discussed include: model-based validation approaches and matching techniques for automotive vision based pedestrian detection;3D face recognition using mapped depth image...
ISBN:
(纸本)0769526608
The proceedings contain 162 papers. The topics discussed include: model-based validation approaches and matching techniques for automotive vision based pedestrian detection;3D face recognition using mapped depth images;a combinational approach to the fusion, de-noising and enhancement of dual-energy x-ray luggage images;multiperspective thermal IR and video arrays for 3D body tracking and driver activity analysis;tracking humans using multi-modal fusion;improved likelihood function in particle-based IR eye tracking;spaceborne traffic monitoring with dual channel synthetic aperture radar – theory and experiments;comparative image fusion analysis;performance evaluation of face recognition using visual and thermal imagery with advanced correlation filters;and integrating LDV audio and ir video for remote multimodal surveillance.
Conventional stereo matching algorithms assume color constancy on the corresponding opaque pixels in the stereo images. However, when the foreground objects with fractional boundary are blended to the scene behind usi...
详细信息
In this paper, we propose a novel online learning method which can learn appearance models incrementally from a given video stream. The data of each frame in the video can be discarded as soon as it has been processed...
详细信息
ISBN:
(纸本)9781424411795
In this paper, we propose a novel online learning method which can learn appearance models incrementally from a given video stream. The data of each frame in the video can be discarded as soon as it has been processed. We only need to maintain a few linear eigenspace models and a transition matrix to approximately construct face appearance manifolds. It is convenient to use these learnt models for video-based face recognition. There are mainly two contributions in this paper First, we propose an algorithm which can learn appearance models online without using a pretrained model. Second, we propose a methodfor eigenspace splitting to prevent that most samples cluster into the same eigenspace. This is useful for clustering and classification. Experimental results show that the proposed method can both learn appearance models online and achieve high recognition rate.
We study the challenging problem of maneuvering object tracking with unknown dynamics, i.e., forces or torque. We investigate the underlying causes of object kinematics, and propose a generative model approach that en...
详细信息
暂无评论