检索结果-内蒙古大学图书馆

30th IEEE Conference on Visual Communications and Image Processing (VCIP)

作者： Mao, Qi Wang, Shanshe Su, Jing Zhang, Xiang Zhang, Xinfeng Ma, Siwei Peking Univ Inst Digital Media Beijing Peoples R China Peking Univ Shenzhen Grad Sch Shenzhen Peoples R China Nanyang Technol Univ Rapid Rich Object Search ROSE Lab Singapore Singapore

ISBN: (纸本)9781509053162

In the 3D extension of Audio Video Coding Standard (AVS), i.e. 3D-AVS, the Global Disparity Vector (GDV) derivation technique has been proposed to provide an estimation for Disparity Vector (DV) in inter-view prediction, where the GDV is generated by averaging all Disparity Vectors (DVs) in the latest previously coded frame. The prediction accuracy of GDV may be however limited by the lack of local adaptivity. In this paper, we introduce a novel Local Disparity Vector (LDV) derivation scheme. Specifically, the DV of the current block is calculated from the DVs within a neighboring region, whose size can be adaptively expanded to increase the robustness and accuracy. Experimental results show that the proposed LDV derivation method can provide around 2.12% and 1.37% bitrate reductions for compressed views and synthesized views compared with the GDV scheme, respectively.

关键词： 3D video coding local disparity vector 3D-AVS

来源：评论

学校读者我要写书评

暂无评论

Joint learning for image-based handbag recommendation

Joint learning for image-based handbag recommendation

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Yan Wang Sheng Li Alex C. Kot Rapid-Rich Object Search (ROSE) Lab Nanyang Technological University Singapore

Fashion recommendation helps shoppers to find desirable fashion items, which facilitates online interaction and product promotion. In this paper, we propose a method to recommend handbags to each shopper, based on the handbag images the shopper has clicked. This is performed by Joint learning of attribute Projection and One-class SVM classification (JPO) based on the images of the shopper's preferred handbags. More specifically, for the handbag images clicked by each shopper, we project the original image feature space into an attribute space which is more compact. The projection matrix is learned jointly with a one-class SVM to yield a shopper-specific one-class classifier. The results show that the proposed JPO handbag recommendation performs favorably based on initial subject testing.

关键词： Support vector machines Feature extraction Joints Testing Recommender systems Data models Collaboration

来源：评论

学校读者我要写书评

暂无评论

IMPROVING object DETECTION WITH REGION SIMILARITY LEARNING

IMPROVING OBJECT DETECTION WITH REGION SIMILARITY LEARNING

引用

IEEE International Conference on Multimedia and Expo (ICME)

作者： Gao, Feng Lou, Yihang Bai, Yan Wang, Shiqi Huang, Tiejun Duan, Ling-Yu Peking Univ Natl Engn Lab Video Technol Beijing Peoples R China Peking Univ Shenzhen Grad Sch SECE Shenzhen Peoples R China Nanyang Technol Univ Rapid Rich Object Search Lab Singapore Singapore

ISBN: (纸本)9781509060672

object detection aims to identify instances of semantic objects of a certain class in images or videos. The success of state-of-the-art approaches is attributed to the significant progress of object proposal and convolutional neural networks (CNNs). Most promising detectors involve multi-task learning with an optimization objective of softmax loss and regression loss. The first is for multi-class categorization, while the latter is for improving localization accuracy. However, few of them attempt to further investigate the hardness of distinguishing different sorts of distracting background regions (i.e., negatives) from true object regions (i.e., positives). To improve the performance of classifying positive object regions vs. a variety of negative background regions, we propose to incorporate triplet embedding into learning objective. The triplet units are formed by assigning each negative region to a meaningful object class and establishing class-specific negatives, followed by triplets construction. Over the benchmark PASCAL VOC 2007, the proposed triplet embedding has improved the performance of well-known Fast-RCNN model with a mAP gain of 2.1%. In particular, the state-of-the-art approach OHEM can benefit from the triplet embedding and has achieved a mAP improvement of 1.2%.

关键词： object detection Similarity distance learning Triplet embedding Region proposal

来源：评论

学校读者我要写书评

暂无评论

Tagging the shoe images by semantic attributes

Tagging the shoe images by semantic attributes

引用

International Conference on Digital Signal Processing (DSP)

作者： Huijing Zhan Sheng Li Alex C. Kot Rapid-Rich Object Search(ROSE) Lab Nanyang Technological University Singapore

ISBN: (纸本)9781479980598

With the rapid proliferation of Internet, it becomes a great challenge to annotate explosive number of objects manually. Especially for the fashion domain where a massive collection of new products come up everyday. Therefore, to save human labor, it is essential to develop an automatic tagging system for those fashion products in a variety of appearances. In this paper, we focus on addressing the issue of automatic shoe tagging where a novel system is proposed to predict the semantic attributes of the shoe images. Given a shoe image in an unknown viewpoint, our proposed system first classify it into one of the 6 pre-defined representative viewpoints, which are commonly displayed in online merchants. To localize the shoe parts on the identified viewpoint, a view-specific part localization model is proposed based on the prior knowledge of the shoe structures under different viewpoints. Finally, we extract several complementary low-level features from the localized shoe parts, which is fed into a SVM classifier for attribute prediction. The effectiveness of the proposed system is demonstrated on a newly-built pump shoe image dataset.

关键词： Footwear Feature extraction Tagging Accuracy Semantics Conferences Support vector machines

来源：评论

学校读者我要写书评

暂无评论

Unimodal and Crossmodal Refinement Network for Multimodal Sequence Fusion

Unimodal and Crossmodal Refinement Network for Multimodal Se...

引用

Conference on Empirical Methods in Natural Language Processing (EMNLP)

作者： Guo, Xiaobao Kong, Adams Zhou, Huan Wang, Xianfeng Wang, Min Nanyang Technol Univ Sch Comp Sci & Engn Singapore Singapore Nanyang Technol Univ Rapid Rich Object Search ROSE Lab IGP Singapore Singapore Huawei Technol AI Applicat Res Ctr AARC Shenzhen Guangdong Peoples R China

ISBN: (纸本)9781955917094

Effective unimodal representation and complementary crossmodal representation fusion are both important in multimodal representation learning. Prior works often modulate one modal feature to another straightforwardly and thus, underutilizing both unimodal and crossmodal representation refinements, which incurs a bottleneck of performance improvement. In this paper, Unimodal and Crossmodal Refinement Network (UCRN) is proposed to enhance both unimodal and crossmodal representations. Specifically, to improve unimodal representations, a unimodal refinement module is designed to refine modality-specific learning via iteratively updating the distribution with transformer-based attention layers. Self-quality improvement layers are followed to generate the desired weighted representations progressively. Subsequently, those unimodal representations are projected into a common latent space, regularized by a multimodal Jensen-Shannon divergence loss for better crossmodal refinement. Lastly, a cross-modal refinement module is employed to integrate all information. By hierarchical explorations on unimodal, bimodal, and trimodal interactions, UCRN is highly robust against missing modality and noisy data. Experimental results on MOSI and MOSEI datasets illustrated that the proposed UCRN outperforms recent state-of-the-art techniques and its robustness is highly preferred in real multimodal sequence fusion scenarios. Codes will be shared publicly(1).

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Aligned Divergent Pathways for Omni-Domain Generalized Person Re-Identification

Aligned Divergent Pathways for Omni-Domain Generalized Perso...

引用

Electrical, Computer and Energy Technologies (ICECET), International Conference on

作者： Eugene P.W. Ang Shan Lin Alex C. Kot Rapid-Rich Object Search (ROSE) Lab Nanyang Technological University Singapore Singapore

ISBN: (数字)9798350395914

ISBN: (纸本)9798350395921

Person Re-identification (Person ReID) has advanced significantly in fully supervised and domain generalized Person R $e$ ID. However, methods developed for one task domain transfer poorly to the other. An ideal Person R $e$ ID method should be effective regardless of the number of domains involved in training or testing. Furthermore, given training data from the target domain, it should perform at least as well as state-of-the-art (SOTA) fully supervised Person ReID methods. We call this paradigm Omni-Domain Generalization Person ReID, referred to as ODG-ReID, and propose a way to achieve this by expanding compatible backbone architectures into multiple diverse pathways. Our method, Aligned Divergent Pathways (ADP), first converts a base architecture into a multi-branch structure by copying the tail of the original backbone. We design our module Dynamic Max-Deviance Adaptive Instance Normalization (DyMAIN) that encourages learning of generalized features that are robust to omni-domain directions and apply DyMAIN to the branches of ADP. Our proposed Phased Mixture-of-Cosines (PMoC) coordinates a mix of stable and turbulent learning rate schedules among branches for further diversified learning. Finally, we realign the feature space between branches with our proposed Dimensional Consistency Metric Loss (DCML). ADP outperforms the state-of-the-art (SOTA) results for multi-source domain generalization and supervised ReID within the same domain. Furthermore, our method demonstrates improvement on a wide range of single-source domain generalization benchmarks, achieving Omni-Domain Generalization over Person ReID tasks.

关键词： Training Measurement Degradation Schedules Training data Tail Computer architecture Dynamic scheduling Standards Identification of persons

来源：评论

学校读者我要写书评

暂无评论

High-Efficiency Image Coding via Near-Optimal Filtering

引用

IEEE SIGNAL PROCESSING LETTERS 2017年第9期24卷 1403-1407页

作者： Zhang, Xinfeng Wang, Shiqi Zhang, Yabin Lin, Weisi Ma, Siwei Gao, Wen Nanyang Technol Univ Rapid Rich Object Search Lab Singapore 639798 Singapore City Univ Hong Kong Dept Comp Sci Kowloon Hong Kong Peoples R China Peking Univ Sch Elect Engn & Comp Sci Inst Digital Media Beijing 100871 Peoples R China

Wiener filtering, which has been widely used in the field of image restoration, is statistically optimal in the sense of mean square error. The adaptive loop filter in video coding inherits the design of Wiener filters, and has been proved to achieve significant improvement on compression performance by reducing coding artifacts and providing high-quality references for subsequent frames. To further improve the compression performance via filtering technique, we explore the factors that may hinder the potential performance of Wiener-based filters, and propose a near-optimal filter learning scheme for high-efficiency image coding. Based on the analyses, we observe that the foremost factor affecting the performance of Wiener-based filters is the divergence of statistical characteristics of training samples, instead of the filter taps or shapes. In view of this, we propose an iterative training method to derive the near-optimal Wiener filter parameters by simultaneously labeling sample categories at the pixel level. These parameters are compressed and transmitted to the decoder side to improve the quality of decoded images by reducing the coding artifacts. Experimental results show that the proposed scheme achieves significant bitrate savings compared with high-efficiency video coding in high-bitrate intra coding scenario.

关键词： High-efficiency video coding (HEVC) image compression near-optimal filter sample classification wiener filter

来源：评论

学校读者我要写书评

暂无评论

Sparsity-Based Image Error Concealment via Adaptive Dual Dictionary Learning and Regularization

引用

IEEE TRANSACTIONS ON IMAGE PROCESSING 2017年第2期26卷 782-796页

作者： Liu, Xianming Zhai, Deming Zhou, Jiantao Wang, Shiqi Zhao, Debin Gao, Huijun Harbin Inst Technol Sch Comp Sci & Technol Harbin 150001 Peoples R China Univ Macau Fac Sci & Technol Macau 999078 Peoples R China Nanyang Technol Univ Rapid Rich Object Search Lab Singapore 639798 Singapore Harbin Inst Technol Sch Astronaut Harbin 150001 Peoples R China

In this paper, we propose a novel sparsity-based image error concealment (EC) algorithm through adaptive dual dictionary learning and regularization. We define two feature spaces: the observed space and the latent space, corresponding to the available regions and the missing regions of image under test, respectively. We learn adaptive and complete dictionaries individually for each space, where the training data are collected via an adaptive template matching mechanism. Based on the piecewise stationarity of natural images, a local correlation model is learned to bridge the sparse representations of the aforementioned dual spaces, allowing us to transfer the knowledge of the available regions to the missing regions for EC purpose. Eventually, the EC task is formulated as a unified optimization problem, where the sparsity of both spaces and the learned correlation model are incorporated. Experimental results show that the proposed method outperforms the state-of-the-art techniques in terms of both objective and perceptual metrics.

关键词： Image error concealment dual dictionary learning sparse coding kernel ridge regression

来源：评论

学校读者我要写书评

暂无评论

Deep Hashing for Scalable Image search

引用

IEEE TRANSACTIONS ON IMAGE PROCESSING 2017年第5期26卷 2352-2367页

作者： Lu, Jiwen Liong, Venice Erin Zhou, Jie Tsinghua Univ Dept Automat Beijing 100084 Peoples R China Tsinghua Univ State Key Lab Intelligent Technol & Syst Beijing 100084 Peoples R China Tsinghua Univ Tsinghua Natl Lab Informat Sci & Technol Beijing 100084 Peoples R China Nanyang Technol Univ Rapid Rich Object Search Lab Interdisciplinary Grad Sch Singapore 639798 Singapore

In this paper, we propose a new deep hashing (DH) approach to learn compact binary codes for scalable image search. Unlike most existing binary codes learning methods, which usually seek a single linear projection to map each sample into a binary feature vector, we develop a deep neural network to seek multiple hierarchical non-linear transformations to learn these binary codes, so that the non-linear relationship of samples can be well exploited. Our model is learned under three constraints at the top layer of the developed deep network: 1) the loss between the compact real-valued code and the learned binary vector is minimized, 2) the binary codes distribute evenly on each bit, and 3) different bits are as independent as possible. To further improve the discriminative power of the learned binary codes, we extend DH into supervised DH (SDH) and multi-label SDH by including a discriminative term into the objective function of DH, which simultaneously maximizes the inter-class variations and minimizes the intra-class variations of the learned binary codes with the single-label and multi-label settings, respectively. Extensive experimental results on eight widely used image search data sets show that our proposed methods achieve very competitive results with the state-of-the-arts.

关键词： Scalable image search fast similarity search hashing deep learning multi-label learning

来源：评论

学校读者我要写书评

暂无评论

Learning-Based Parallax Transfer on Multispectral Light Field 23

Learning-Based Parallax Transfer on Multispectral Light Fiel...

引用

23rd IEEE International Conference on Digital Signal Processing (DSP)

作者： Nan, Shengyu Chen, Jie Chau, Lap-Pui Qian, Kemao Nanyang Technol Univ Interdisciplinary Grad Sch Rapid Rich Object Search Lab Singapore Singapore Nanyang Technol Univ Sch Elect & Elect Engn Singapore Singapore Nanyang Technol Univ Sch Comp Sci & Engn Singapore Singapore

ISBN: (纸本)9781538668122;9781538668115

In light field imaging, one challenging task is to fuse the spatial and angular information with spectral information. Existing methods for multi-modal and multi-spectral correspondence problem are descriptors based algorithms which are time consuming and not able to handle the small base line issue in light field. In recent years, deep learning based methods demonstrate good performance in many challenging computer vision areas. Inspired by their powerful performance, we propose a learning based method to transfer the parallax information across channels in light field. We exploit spatial and angular information from two reference channels and the spatial information from the target channel to predict the different views and finally reconstruct the target channel. Experimental results demonstrate that compared with other descriptors based methods, our learning based method is much less time consuming and able to effectively transfer the parallax information even when the parallax shift is very small.

关键词： light field multispectral parallax transfer deep learning

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：