Social brainstorming has become a popular group creativity technique to create a large number of ideas for solving problems. The social computing, for its advance in information processing and network communication, c...
详细信息
Most previous research on image categorization has focused on medium-scale data sets, while large-scale image categorization with millions of images from thousands of categories remains a challenge. With the emergence...
详细信息
ISBN:
(纸本)9781618395993
Most previous research on image categorization has focused on medium-scale data sets, while large-scale image categorization with millions of images from thousands of categories remains a challenge. With the emergence of structured large-scale dataset such as the imageNet, rich information about the conceptual relationships between images, such as a tree hierarchy among various image categories, become available. As human cognition of complex visual world benefits from underlying semantic relationships between object classes, we believe a machine learning system can and should leverage such information as well for better performance. In this paper, we employ such semantic relatedness among image categories for large-scale image categorization. Specifically, a category hierarchy is utilized to properly define loss function and select common set of features for related categories. An efficient optimization method based on proximal approximation and accelerated parallel gradient method is introduced. Experimental results on a subset of imageNet containing 1.2 million images from 1000 categories demonstrate the effectiveness and promise of our proposed approach.
This 4-Volume-Set, CCIS 0251 - CCIS 0254, constitutes the refereed proceedings of the International conference on Informatics Engineering and Information Science, ICIEIS 2011, held in Kuala Lumpur, Malaysia, in Novemb...
ISBN:
(数字)9783642253270
ISBN:
(纸本)9783642253263
This 4-Volume-Set, CCIS 0251 - CCIS 0254, constitutes the refereed proceedings of the International conference on Informatics Engineering and Information Science, ICIEIS 2011, held in Kuala Lumpur, Malaysia, in November 2011. The 210 revised full papers presented together with invited papers in the 4 volumes were carefully reviewed and selected from numerous submissions. The papers are organized in topical sections on e-learning, information security, software engineering, imageprocessing, algorithms, artificial intelligence and soft computing, e-commerce, data mining, neural networks, social networks, grid computing, biometric technologies, networks, distributed and parallel computing, wireless networks, information and data management, web applications and software systems, multimedia, ad hoc networks, mobile computing, as well as miscellaneous topics in digital information and communications.
This 4-Volume-Set, CCIS 0251 - CCIS 0254, constitutes the refereed proceedings of the International conference on Informatics Engineering and Information Science, ICIEIS 2011, held in Kuala Lumpur, Malaysia, in Novemb...
ISBN:
(数字)9783642254833
ISBN:
(纸本)9783642254826
This 4-Volume-Set, CCIS 0251 - CCIS 0254, constitutes the refereed proceedings of the International conference on Informatics Engineering and Information Science, ICIEIS 2011, held in Kuala Lumpur, Malaysia, in November 2011. The 210 revised full papers presented together with invited papers in the 4 volumes were carefully reviewed and selected from numerous submissions. The papers are organized in topical sections on e-learning, information security, software engineering, imageprocessing, algorithms, artificial intelligence and soft computing, e-commerce, data mining, neural networks, social networks, grid computing, biometric technologies, networks, distributed and parallel computing, wireless networks, information and data management, web applications and software systems, multimedia, ad hoc networks, mobile computing, as well as miscellaneous topics in digital information and communications.
The edge-directed interpolation scheme is a non-iterative, orientation-adaptive method to enhance image resolution with better visual effect than conventional interpolation methods. It interpolates the missing pixels ...
详细信息
The edge-directed interpolation scheme is a non-iterative, orientation-adaptive method to enhance image resolution with better visual effect than conventional interpolation methods. It interpolates the missing pixels based on the covariance of a high-resolution image estimated from the covariance of the low-resolution image. In spite of the impressive performance, the computational complexity of covariance-based adaptation is significantly higher than that of the conventional linear interpolation algorithms. In this paper, we propose a GPU-based massively parallel version of the edge-directed interpolation scheme. A speedup of 61.7x can be achieved with respect to its single-threaded CPU counterpart in the host computer.
In the present paper we are introducing a new method of salient object detection with very good results relative to other already known segmentation methods. We address through our research the problem of image segmen...
详细信息
ISBN:
(纸本)9781601321916
In the present paper we are introducing a new method of salient object detection with very good results relative to other already known segmentation methods. We address through our research the problem of image segmentation evaluation by an efficient comparison of four complex algorithms. In order to compare our method with other approaches, we built an evaluation framework that helped us with our experiments. The experimental results offer a complete basis for parallel analysis with respect to the precision of our algorithm, rather than the individual efficiency.
In the modern world, surveillance is ubiquitous and pervasive. One of the most common such surveillance is video surveillance. While the present day technological advances have provided the state-of-the-art gadgetry f...
详细信息
With the rapid expansion of Internet and distributed technology, multimedia security and digital rights management have been received much attention in the literature, As a major method for intellectual property right...
详细信息
Recent advances in digital photography have enabled the development and demonstration of plenoptic cameras with impressive capabilities. They function by recording sub-aperture images that can be combined to re-focus ...
详细信息
ISBN:
(纸本)9780819485946
Recent advances in digital photography have enabled the development and demonstration of plenoptic cameras with impressive capabilities. They function by recording sub-aperture images that can be combined to re-focus images or to generate stereoscopic pairs. Plenoptic methods are being explored for fusing images from distributed arrays of cameras, with a view toward applications in which hardware resources are limited (e. g. size, weight, power constraints). Through computer simulation and experimental studies, the influences of non-idealities such as camera position uncertainty are being considered. Component image rescaling and balancing methods are being explored to compensate. Of interest is the impact on precision passive ranging and super-resolution. In a preliminary experiment, a set of images from a camera array was recorded and merged to form a 3D representation of a scene. Conventional plenoptic refocusing was demonstrated and techniques were explored for balancing the images. Nonlinear methods were explored for combining the images limited the ghosting caused by sub-sampling. Plenoptic processing was explored as a means for determining 3D information from airborne video. Successive frames were processed as camera array elements to extract the heights of structures. Practical means were considered for rendering the 3D information in color.
High efficient acquisition of the sensor array signals and accurate reconstruction of the backscatters medium are important issues in ultrasound reveive beamforming. This paper presents a novel receive beamforming app...
详细信息
暂无评论