In this paper, we present a new hashing method to learn compact binary codes for highly efficient image retrieval on large-scale datasets. While the complex image appearance variations still pose a great challenge to ...
详细信息
ISBN:
(纸本)9781467388528
In this paper, we present a new hashing method to learn compact binary codes for highly efficient image retrieval on large-scale datasets. While the complex image appearance variations still pose a great challenge to reliable retrieval, in light of the recent progress of Convolutional Neural Networks (CNNs) in learning robust image representation on various vision tasks, this paper proposes a novel Deep Supervised Hashing (DSH) method to learn compact similarity-preserving binary code for the huge body of image data. Specifically, we devise a CNN architecture that takes pairs of images (similar/dissimilar) as training inputs and encourages the output of each image to approximate discrete values (e.g. +1/-1). To this end, a loss function is elaborately designed to maximize the discriminability of the output space by encoding the supervised information from the input image pairs, and simultaneously imposing regularization on the real-valued outputs to approximate the desired discrete values. For image retrieval, new-coming query images can be easily encoded by propagating through the network and then quantizing the network outputs to binary codes representation. Extensive experiments on two large scale datasets CIFAR-10 and NUS-WIDE show the promising performance of our method compared with the state-of-the-arts.
The existed cloud data models can not well meet the management requirements of structured data including a great deal of relational data,therefore a two-layer cloud data model consisting of a presentation layer and a ...
详细信息
The existed cloud data models can not well meet the management requirements of structured data including a great deal of relational data,therefore a two-layer cloud data model consisting of a presentation layer and a storage layer is *** the presentation layer,the conception of composite class and composite object are defined to represent the structure and the data of structured data *** the storage layer,a composite class in the presentation layer is transformed into a3-tuple which preserves data structure in mappings between composite attributes and their sub-attributes by using a rule set;and a composite object in the presentation layer is transformed into another 3-tuple which preserves a structured data set in simple objects without embed objects by using a CAO(Component-Attribute-Object) set in which each element mainly consists of a simple object and the identification of its father *** order to store data in key-value model,a method to convert a CAO to a pair of key-value is ***,two algorithms were proposed to convert data between the representation layer and the storage *** experiment shows the two conversion algorithms are *** proposed model is fit for the management requirements of structured data in the cloud because composite object can represent structured data and avoid join operation of relation data and CAO can be stored in key-value model.
Object recognition has many applications in human-machine interaction and multimedia retrieval. However, due to large intra-class variability and inter-class similarity, accurate recognition relying only on RGB data i...
详细信息
Object recognition has many applications in human-machine interaction and multimedia retrieval. However, due to large intra-class variability and inter-class similarity, accurate recognition relying only on RGB data is still a big challenge. Recently, with the emergence of inexpensive RGB-D devices, this challenge can be better addressed by leveraging additional depth information. A very special yet important case of object recognition is hand-held object recognition, as manipulating objects with hands is common and intuitive in human-human and human-machine interactions. In this paper, we study this problem and introduce an effective framework to address it. This framework first detects and segments the hand-held object by exploiting skeleton information combined with depth information. In the object recognition stage, this work exploits heterogeneous features extracted from different modalities and fuses them to improve the recognition accuracy. In particular, we incorporate handcrafted and deep learned features and study several multi-step fusion variants. Experimental evaluations validate the effectiveness of the proposed method.
The context of objects can provide auxiliary discrimination beyond objects. However, this effective information has not been fully explored. In this paper, we propose Tri-level Combination for Image Representation (Tr...
详细信息
Protein folding is one of the most important problems in molecular biology. The kinetic order of protein folding is one of the main aspects of the folding process. Previous methods for predicting protein folding kinet...
详细信息
In this Letter, we propose a color holographic zoom system based on a liquid lens. We use the spatial multiplexing method to realize color reconstruction. By controlling the focal lengths of the liquid lens and the en...
详细信息
In this Letter, we propose a color holographic zoom system based on a liquid lens. We use the spatial multiplexing method to realize color reconstruction. By controlling the focal lengths of the liquid lens and the encoded digital lens on the spatial light modulator panel, we can change the magnification of the reconstructed image very quickly, without mechanical parts and keeping the output plane stationary.
Threshold secret sharing (SS), also denoted as (t, n) SS, has been used extensively in the area of information se- curity, such as for group authentication, cloud storage schemes, secure parallel communication and wir...
详细信息
Threshold secret sharing (SS), also denoted as (t, n) SS, has been used extensively in the area of information se- curity, such as for group authentication, cloud storage schemes, secure parallel communication and wireless mul- tipath routing protocols. However, a (t, n) SS cannot de- tect any deceptions among the dealer and shareholders. Veriable secret sharing (VSS) overcomes the weakness of (t, n) SS in such a way that it is able to detect cheaters by verifying the validity of shares or the correctness of the recovered secret under the condition that both shares and the secret are not compromised. Recently, two non- interactive VSSs based on Asmuth-Bloom's SS were pro- posed by Harn et al. and Liu et al., respectively. Both VSSs require shareholders to examine the range of values of some integers related to the secret before recovering the secret, which is a time-consuming operation. In this paper, we propose a novel integratable VSS mechanism that integrates the concepts of the generalized Chinese remainder theorem (GCRT), Shamir's SS and Asmuth- Bloom's SS. Our proposed VSS can verify that the secret reconstructed by any t or more shareholders is the same as the one that the dealer has generated. Analysis shows that our proposed VSS can provide perfect secrecy and better efficiency.
Owing to the shortages of inconvenience, expensive and high professional requirements etc. for conventional recognition devices of wheat leaf diseases, it does not satisfy the requirements of uploadin
Owing to the shortages of inconvenience, expensive and high professional requirements etc. for conventional recognition devices of wheat leaf diseases, it does not satisfy the requirements of uploadin
Cross-modal hashing has received more and more attention because of its fast query speed and low storage cost. In this paper, we propose a flexible yet simple cross-modal hashing method to deal with the problem of cro...
详细信息
ISBN:
(纸本)9781467384155
Cross-modal hashing has received more and more attention because of its fast query speed and low storage cost. In this paper, we propose a flexible yet simple cross-modal hashing method to deal with the problem of cross-modal retrieval. The proposed method consists of two steps. In the first phase, we use a kernel canonical correlation analysis method named Anchor kernel canonical correlation analysis (AKCCA) to map data from different modalities into a common kernel space. In the second phase, we use the method named Supervised Hashing with Kernels (KSH) to learn hashing functions bit by bit. These two useful ingredients are combined seamlessly to achieve promising results. Experimental results on a benchmark dataset demonstrate that our method performs better than several state-of-the-art methods.
The OWA operator is an important assessment method in multiple-attribute decision making. A new model based on the density function of normal distribution is given to assign reasonable weights of OWA operators. Using ...
详细信息
The OWA operator is an important assessment method in multiple-attribute decision making. A new model based on the density function of normal distribution is given to assign reasonable weights of OWA operators. Using the orness level as a parameter in this method, one can deduce OWA weights by solving a quadratic programing problem. Three propositions of this new model are proven, and a numerical example about the assessment of red wines is given to analyze and illustrate this method.
暂无评论