检索结果-内蒙古大学图书馆

State-relabeling adversarial active learning

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhang, Beichen Li, Liang Yang, Shijie Wang, Shuhui Zha, Zheng-Jun Huang, Qingming University of Chinese Academy of Sciences Beijing China Key Lab of Intell. Info. Process. Inst. of Comput. Tech. Chinese Academy of Sciences China University of Science and Technology of China China Peng Cheng Laboratory Shenzhen China

Active learning is to design label-efficient algorithms by sampling the most representative samples to be labeled by an oracle. In this paper, we propose a state relabeling adversarial active learning model (SRAAL), that leverages both the annotation and the labeled/unlabeled state information for deriving the most informative unlabeled samples. The SRAAL consists of a representation generator and a state discriminator. The generator uses the complementary annotation information with traditional reconstruction information to generate the unified representation of samples, which embeds the semantic into the whole data representation. Then, we design an online uncertainty indicator in the discriminator, which endues unlabeled samples with different importance. As a result, we can select the most informative samples based on the discriminator’s predicted state. We also design an algorithm to initialize the labeled pool, which makes subsequent sampling more efficient. The experiments conducted on various datasets show that our model outperforms the previous state-of-art active learning methods and our initially sampling algorithm achieves better performance. Copyright © 2020, The Authors. All rights reserved.

关键词： Semantics

Parsing-based view-aware embedding network for vehicle re-identification

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Meng, Dechao Li, Liang Liu, Xuejing Li, Yadong Yang, Shijie Zha, Zheng-Jun Gao, Xingyu Wang, Shuhui Huang, Qingming Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing China University of Chinese Academy of Sciences China Megvii Inc Beijing China University of Science and Technology of China China Peng Cheng Laboratory Shenzhen China Institute of Microelectronics Chinese Academy of Sciences Beijing China

Vehicle Re-Identification is to find images of the same vehicle from various views in the cross-camera scenario. The main challenges of this task are the large intra-instance distance caused by different views and the subtle inter-instance discrepancy caused by similar vehicles. In this paper, we propose a parsing-based view-aware embedding network (PVEN) to achieve the view-aware feature alignment and enhancement for vehicle ReID. First, we introduce a parsing network to parse a vehicle into four different views, and then align the features by mask average pooling. Such alignment provides a fine-grained representation of the vehicle. Second, in order to enhance the view-aware features, we design a common-visible attention to focus on the common visible views, which not only shortens the distance among intra-instances, but also enlarges the discrepancy of inter-instances. The PVEN helps capture the stable discriminative information of vehicle under different views. The experiments conducted on three datasets show that our model outperforms state-of-the-art methods by a large margin. Copyright © 2020, The Authors. All rights reserved.

关键词： Vehicles

Parsing-Based View-Aware Embedding Network for Vehicle Re-Identification

学校读者我要写书评

暂无评论

Parsing-Based View-Aware Embedding Network for Vehicle Re-Id...

Conference on comput.r Vision and Pattern Recognition (CVPR)

作者： Dechao Meng Liang Li Xuejing Liu Yadong Li Shijie Yang Zheng-Jun Zha Xingyu Gao Shuhui Wang Qingming Huang Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing China University of Chinese Academy of Sciences China Megvii Inc Beijing China University of Science and Technology of China China Institute of Microelectronics Chinese Academy of Sciences Beijing China Peng Cheng Laboratory Shenzhen China

ISBN: (数字)9781728171685

ISBN: (纸本)9781728171692

Vehicle Re-Identification is to find images of the same vehicle from various views in the cross-camera scenario. The main challenges of this task are the large intra-instance distance caused by different views and the subtle inter-instance discrepancy caused by similar vehicles. In this paper, we propose a parsing-based view-aware embedding network (PVEN) to achieve the view-aware feature alignment and enhancement for vehicle ReID. First, we introduce a parsing network to parse a vehicle into four different views and then align the features by mask average pooling. Such alignment provides a fine-grained representation of the vehicle. Second, in order to enhance the view-aware features, we design a common-visible attention to focus on the common visible views, which not only shortens the distance among intra-instances, but also enlarges the discrepancy of inter-instances. The PVEN helps capture the stable discriminative information of vehicle under different views. The experiments conducted on three datasets show that our model outperforms state-of-the-art methods by a large margin.

关键词： Feature extraction Task analysis Cameras Gallium nitride Training Image color analysis Fuses

Stacked Cross Refinement Network for Edge-Aware Salient Object Detection

学校读者我要写书评

暂无评论

Stacked Cross Refinement Network for Edge-Aware Salient Obje...

International Conference on comput.r Vision (ICCV)

作者： Zhe Wu Li Su Qingming Huang School of Computer Science and Technology University of Chinese Academy of Sciences (UCAS) Beijing China Key Lab of Big Data Mining and Knowledge Management UCAS Beijing China Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing China Peng Cheng Laboratory ShenZhen China

ISBN: (数字)9781728148038

ISBN: (纸本)9781728148045

Salient object detection is a fundamental comput.r vision task. The majority of existing algorithms focus on aggregating multi-level features of pre-trained convolutional neural networks. Moreover, some researchers attempt to utilize edge information for auxiliary training. However, existing edge-aware models design unidirectional frameworks which only use edge features to improve the segmentation features. Motivated by the logical interrelations between binary segmentation and edge maps, we propose a novel Stacked Cross Refinement Network (SCRN) for salient object detection in this paper. Our framework aims to simultaneously refine multi-level features of salient object detection and edge detection by stacking Cross Refinement Unit (CRU). According to the logical interrelations, the CRU designs two direction-specific integration operations, and bidirectionally passes messages between the two tasks. Incorporating the refined edge-preserving features with the typical U-Net, our model detects salient objects accurately. Extensive experiments conducted on six benchmark datasets demonstrate that our method outperforms existing state-of-the-art algorithms in both accuracy and efficiency. Besides, the attribute-based performance on the SOC dataset show that the proposed model ranks first in the majority of challenging scenes. Code can be found at https://***/wuzhe71/SCAN.

关键词： Image edge detection Task analysis Feature extraction Object detection Image segmentation Silicon comput.tional modeling

Learning personalized attribute preference via multi-task auc optimization

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Yang, Zhiyong Xu, Qianqian Cao, Xiaochun Huang, Qingming SKLOIS Institute of Information Engineering Chinese Academy of Sciences Beijing China School of Cyber Security University of Chinese Academy of Sciences Beijing China Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing China University of Chinese Academy of Sciences Beijing China Key Laboratory of Big Data Mining and Knowledge Management CAS Beijing China

Traditionally, most of the existing attribute learning methods are trained based on the consensus of annotations aggregated from a limited number of annotators. However, the consensus might fail in settings, especially when a wide spectrum of annotators with different interests and comprehension about the attribute words are involved. In this paper, we develop a novel multi-task method to understand and predict personalized attribute annotations. Regarding the attribute preference learning for each annotator as a specific task, we first propose a multi-level task parameter decomposition to capture the evolution from a highly popular opinion of the mass to highly personalized choices that are special for each person. Meanwhile, for personalized learning methods, ranking prediction is much more important than accurate classification. This motivates us to employ an Area Under ROC Curve (AUC) based loss function to improve our model. On top of the AUC-based loss, we propose an efficient method to evaluate the loss and gradients. Theoretically, we propose a novel closed-form solution for one of our non-convex subproblem, which leads to provable convergence behaviors. Furthermore, we also provide a generalization bound to guarantee a reasonable performance. Finally, empirical analysis consistently speaks to the efficacy of our proposed method. Copyright © 2019, The Authors. All rights reserved.

关键词： Machine learning

From seed discovery to deep reconstruction: Predicting saliency in crowd via deep networks 16

学校读者我要写书评

暂无评论

From seed discovery to deep reconstruction: Predicting salie...

24th ACM Multimedia Conference, MM 2016

作者： Zhang, Yanhao Qin, Lei Huang, Qingming Yang, Kuiyuan Zhang, Jun Yao, Hongxun Institute of Technology Harbin China Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS China Microsoft Research Asia Beijing China Hefei University of Technology Hefei China

ISBN: (纸本)9781450336031

Although saliency prediction in crowd has been recently recognized as an essential task for video analysis, it is not comprehensively explored yet. The challenges lie in that eyefixations in crowded scenes are inherently\distinct"and\multimodal", which difiers from those in regular scenes. To this end, the existing saliency prediction schemes typically rely on hand designed features with shallow learning paradigm, which neglect the underlying characteristics of crowded scenes. In this paper, we propose a saliency prediction model dedicated for crowd videos with two novelties: 1) Distinct units are discovered using deep representation learned by a Stacked Denoising Auto-Encoder (SDAE), considering perceptual properties of crowd saliency;2) Contrast-based saliency is measured through deep reconstruction errors in the second SDAE trained on all units excluding distinct units. A unified model is integrated for online process.ng crowd saliency. Extensive evaluations on two crowd video benchmark datasets demonstrate that our approach can effectively explore crowd saliency mechanism in two-stage SDAEs and achieve significantly better results than state-of-the-art methods, with robustness to parameters. © 2016 ACM.

关键词： Forecasting

Exploring outliers in crowdsourced ranking for QoE

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Xu, Qianqian Yan, Ming Huang, Chendi Xiong, Jiechao Huang, Qingming Yao, Yuan Institute of Information Engineering Cas Beijing100093 Department of Computational Mathematics Michigan State University East LansingMI48824 United States BICMR-LMAM-LMEQF-LMP School of Mathematical Sciences Peking University Beijing100871 Tencent Ai Lab Shenzhen518057 University of Chinese Academy of Sciences Beijing100049 China Key Lab of Intell. Info. Process. Inst. of Comput. Tech. Cas Beijing100190 Key Lab of Big Data Mining and Knowledge Management Cas Beijing100190 Department of Mathematics Hong Kong University of Science and Technology 100871 Hong Kong

Outlier detection is a crucial part of robust evaluation for crowd-sourceable assessment of Quality of Experience (QoE) and has attracted much attention in recent years. In this paper, we propose some simple and fast algorithms for outlier detection and robust QoE evaluation based on the nonconvex optimization principle. Several iterative procedures are designed with or without knowing the number of outliers in samples. Theoretical analysis is given to show that such procedures can reach statistically good estimates under mild conditions. Finally, experimental results with simulated and real-world crowdsourcing datasets show that the proposed algorithms could produce similar performance to Huber-LASSO approach in robust ranking, yet with nearly 8 or 90 times speed-up, without or with a prior knowledge on the sparsity size of outliers, respectively. Therefore the proposed methodology provides us a set of helpful tools for robust QoE evaluation with crowdsourcing data. Copyright © 2017, The Authors. All rights reserved.

关键词： Crowdsourcing

Region annotations in hashing based image retrieval 14

学校读者我要写书评

暂无评论

Region annotations in hashing based image retrieval

6th International Conference on Internet Multimedia comput.ng and Service, ICIMCS 2014

作者： Guo, Cheng Herranz, Luis Wu, Lifang Jiang, Shuqiang Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing 100190 China School of Electronic Information and Control Engineering Beijing University of Technology Beijing China

ISBN: (纸本)9781450328104

During recent years, large-scale data has become more and more present in our daily life, requiring better, faster, and more effective ways to discriminate between unrelated content and what we are truly interested in. In particular, retrieving similar images is a fundamental problem in both image process.ng and comput.r vision. Since finding similar images high-dimensional image features is not feasible in practice, approximate hashing techniques have been proved very effective due to their great efficiency and reasonably accuracy. Current hashing methods focus on image as a whole. However, sometimes the relevant content might be just bounded to a region of interest while the background part seems to be needless for users. In this paper, we propose to exploit region-level annotations, whenever they are available, for both training supervised hashing methods and also in the query image. Our objective is to evaluate whether region-level features can be helpful in a supervised hashing scenario. Experimental results confirm that region-level annotations in both the query and training images increase the accuracy. Copyright 2014 ACM.

关键词： Image retrieval

Multiple feature fusion based hand-held object recognition with RGB-D data 14

学校读者我要写书评

暂无评论

Multiple feature fusion based hand-held object recognition w...

6th International Conference on Internet Multimedia comput.ng and Service, ICIMCS 2014

作者： Liu, Shuang Wang, Shuang Wu, Lifang Jiang, Shuqiang Key Lab of Intell. Info. Process. Inst. of Comput. Tech. CAS Beijing 100190 China School of Electronic Information and Control Engineering Beijing University of Technology Beijing 100124 China

ISBN: (纸本)9781450328104

With the advance of comput.r technology and smart device, many technologies and applications have been developed to enhance the efficiency of human-comput.r interaction (HCI). For human, the hand is a natural and direct way in communication. Hand-held Object Recognition (HHOR), which is to predict the label for the object people hold in hand, can help machines in understanding the environment and people's intentions. However, it has not been well studied in the community. So, in this paper, we proposed a novel feature fusion based method for hand-held object recognition with RGB-D data. First, the skeleton information is used to initially locate the object and with depth map we extract object region in a region-growing manner. Then on the corresponding object point cloud, we use Multiple Kernel Learning (MKL) to fuse the shape feature with color feature to obtain the advantages of them. Specially, we collected a dataset, which contains 12800 video frames of 16 categories and each frame captures the visual image, depth map and user skeleton data. The experiment shows promising results in both segmentation and recognition. Copyright 2014 ACM.

关键词： Object recognition