检索结果-内蒙古大学图书馆

IEEE Transactions on Artificial Intelligence 2021年第1期2卷 58-70页

作者： Li, Ziqiang Tao, Rentuo Wang, Jie Li, Fu Niu, Hongjing Yue, Mingdao Li, Bin The CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application Systems University of Science and Technology of China Hefei230052 China The Department of Electronic Engineering and Information Science University of Science and Technology of China Hefei230052 China The School of Information Science and Technology University of Science and Technology of China Hefei230052 China The College of Mechanical and Electrical Engineering Suzhou University Suzhou234000 China

With the success of generative adversarial networks (GANs) on various real-world applications, the controllability and security of GANs have raised more and more concerns from the community. Specifically, understanding the latent space of GANs, i.e., obtaining the completely decoupled latent space, is essential for applications in some secure scenarios. At present, there is no quantitative method to measure the decoupling of latent space, which is not conducive to the development of the community. In this article, we propose two methods to measure the sensitivity of latent dimensions: one is a sequential intervention method, and the other is an optimization-based method that measures the sensitivity in both the value and the direction. With these two methods, the decoupling of latent space can be measured by the sparsity of the sensitivity vector obtained. The effectiveness of the proposed methods has been verified by experiments on the representative GANs. Code will be available at https://***/iceli1007/latent-analysis-of. Impact Statement-Generative adversarial networks (GANs) is a popular technology in image generation. Benefiting from the development of neural networks, GANs have been widely used in many real-world tasks, such as super-resolution, image translation, and image inpainting. However, the understanding of the generator in GANs as a map from latent space (represented in a multidimensional vector) to image space is incomplete. To make better use of GANs, more insight into the generation process is required. The motivation of this article is to develop methods to analyze the influence of each dimension of latent space on the generated results, and to measure the decoupling of latent space, the knowledge obtained will be useful for the development of more precise controllable image generation technology. © 2021 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Multi-Aspect SAR Target Recognition Based on Efficientnet and GRU

Multi-Aspect SAR Target Recognition Based on Efficientnet an...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Pengfei Zhao Lijia Huang Key Laboratory of Technology in Geo-spatial Information Processing and Application System Chinese of Sciences Beijing China

ISBN: (数字)9781728163741

ISBN: (纸本)9781728163758

Automatic target recognition (ATR) in synthetic aperture radar (SAR) has been extensively applied in military and civilian fields recently. However, SAR images are very sensitive to the azimuth of the imaging, and the same target at different aspects differs greatly, thus requiring more reliable and robust multi-aspect ATR recognition performance. In this paper, we propose an end-to-end multi-aspect ATR model based on EfficientNet and GRU, and use island loss as the training loss, which is more suitable for SAR ATR. Experiments show that our proposed method can achieve 100% accuracy for 10-class recognition, and 99.68% for a large depression angle. Besides, the proposed method can achieve satisfactory accuracy even with reduced datasets. Experimental results have shown that our proposed method outperform other state-of-the-art ATR methods.

关键词： Training Synthetic aperture radar Feature extraction Radar polarimetry Target recognition Data models Image sequences

来源：评论

学校读者我要写书评

暂无评论

Dominant Physical Scattering Mechanism Analysis for GF-3 Typical Ground Objects by Polarimetric Decomposition

Dominant Physical Scattering Mechanism Analysis for GF-3 Typ...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Jin Yan Qiu Xiaolan Lijia Huang Laboratory of Spatial Information Intelligent Processing System Institute of Electronics Chinese Academy of Sciences Suzhou China Key Laboratory of Technology in Geo-spatial Information Processing and Application System CAS

The dominant scattering mechanism is of great significance for the application of ground objects classification and target detection. It can also verify the quality of the polarimetric data by check the dominant scattering mechanism of known ground objects. In order to improve the application performance, this paper studies the dominant scattering mechanism of GF-3 typical ground objects based on a large number of data slices. The GF-3 fully polarimetric data slices are classified based on the MODIS global classification map, and the GF-3 slice library of typical ground objects is constructed. Based on large amounts of GF-3 samples, we carry out the statistical analysis of dominant scattering mechanism separation results for typical GF-3 ground objects (building, woodland, cultivated land, grassland and waters) of by means of h/alpha/A decomposition. The quantitative results reveal the polarimetric scattering feature of different ground objects, and provide reference for fully polarimetric SAR application.

关键词： Scattering Synthetic aperture radar Entropy Buildings Statistical analysis Radar polarimetry Sensors

来源：评论

学校读者我要写书评

暂无评论

Accelerate Neural Image Compression with Channel-Adaptive Arithmetic Coding

Accelerate Neural Image Compression with Channel-Adaptive Ar...

引用

IEEE International Symposium on Circuits and systems

作者： Zongyu Guo Jun Fu Runsen Feng Zhibo Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

We have witnessed the revolutionary progress of learned image compression despite a short history of this field. Some challenges still remain such as computational complexity that prevent the practical application of learning-based codecs. In this paper, we address the issue of heavy time complexity from the view of arithmetic coding. Prevalent learning-based image compression scheme first maps the natural image into latent representations and then conduct arithmetic coding on quantized latent maps. Previous arithmetic coding schemes define the start and end value of the arithmetic codebook as the minimum and maximum of the whole latent maps, ignoring the fact that the value ranges in most channels are shorter. Hence, we propose to use a channel-adaptive codebook to accelerate arithmetic coding. We find that the latent channels have different frequency-related characteristics, which are verified by experiments of neural frequency filtering. Further, the value ranges of latent maps are different across channels which are relatively image-independent. The channel-adaptive characteristics allow us to establish efficient prior codebooks that cover more appropriate ranges to reduce the runtime. Experimental results demonstrate that both the arithmetic encoding and decoding can be accelerated while preserving the rate-distortion performance of compression model.

关键词： Image coding Runtime Filtering Rate-distortion Acceleration Integrated circuit modeling Time complexity

来源：评论

学校读者我要写书评

暂无评论

GUID-based mobile visual communication using NDN mechanism

GUID-based mobile visual communication using NDN mechanism

引用

IEEE Visual Communications and Image processing (VCIP)

作者： Yuanzun Zhang Xiaobin Tan Hao Liu Weiping Li CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

ISBN: (纸本)9781509053179

With the explosive growth in the number of mobile terminals, the demand for visual communication with mobility is increasing. However, traditional solutions for mobility over IP network cannot always meet the demand of satisfying visual communication. Named Data Networking (NDN) is a new communication model that aims to replace IP model brings a different background to mobile visual communication problems. In this paper, we take advantage of the NDN model to realize seamless mobile visual communication. We introduce a delegate with calculation functions and a globally unique identifier (GUID) which can provide native identity indication into the NDN mechanism. The use of GUID benefits real-time applications like visual communication and further works with the delegate to decrease unnecessary routing update. We also specify the naming rule and design a FIB+ to support seamless mobile visual communication. To test the performance of our solutions, we build a proof-of-concept prototype and run experiments on it. The experiments demonstrate that our solution can provide real-time video communication with seamless mobility experience.

关键词： Routing Mobile communication Visual communication IP networks Artificial neural networks Mobile computing Real-time systems

来源：评论

学校读者我要写书评

暂无评论

Learned Video Compression with Feature-level Residuals

Learned Video Compression with Feature-level Residuals

引用

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

作者： Runsen Feng Yaojun Wu Zongyu Guo Zhizheng Zhang Zhibo Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

ISBN: (数字)9781728193601

ISBN: (纸本)9781728193618

In this paper, we present an end-to-end video compression network for P-frame challenge on CLIC. We focus on deep neural network (DNN) based video compression, and improve the current frameworks from three aspects. First, we notice that pixel space residuals is sensitive to the prediction errors of optical flow based motion compensation. To suppress the relative influence, we propose to compress the residuals of image feature rather than the residuals of image pixels. Furthermore, we combine the advantages of both pixel-level and feature-level residual compression methods by model ensembling. Finally, we propose a step-by-step training strategy to improve the training efficiency of the whole framework. Experiment results indicate that our proposed method achieves 0.9968 MS-SSIM on CLIC validation set and 0.9967 MS-SSIM on test set.

关键词： Image coding Training Optical imaging Adaptive optics Motion compensation Optical distortion Decoding

来源：评论

学校读者我要写书评

暂无评论

3-D Context Entropy Model for Improved Practical Image Compression

3-D Context Entropy Model for Improved Practical Image Compr...

引用

IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

作者： Zongyu Guo Yaojun Wu Runsen Feng Zhizheng Zhang Zhibo Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

ISBN: (数字)9781728193601

ISBN: (纸本)9781728193618

In this paper, we present our image compression framework designed for CLIC 2020 competition. Our method is based on Variational AutoEncoder (VAE) architecture which is strengthened with residual structures. In short, we make three noteworthy improvements here. First, we propose a 3-D context entropy model which can take advantage of known latent representation in current spatial locations for better entropy estimation. Second, a light-weighted residual structure is adopted for feature learning during entropy estimation. Finally, an effective training strategy is introduced for practical adaptation with different resolutions. Experiment results indicate our image compression method achieves 0.9775 MS-SSIM on CLIC validation set and 0.9809 MS-SSIM on test set.

关键词： Context modeling Solid modeling Image coding Entropy Training Transforms Image resolution

来源：评论

学校读者我要写书评

暂无评论

Action recognition with novel high-level pose features

Action recognition with novel high-level pose features

引用

IEEE International Conference on Multimedia and Expo Workshops (ICMEW)

作者： Jiayi Fan Zhengjun Zha Xinmei Tian University of Science and Technology of China CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System

ISBN: (纸本)9781509015535

Recently high-level pose features (HLPF) have been shown to be efficient for action recognition in joint-annotated tasks. However, the relative positions between pairs of joints in actual situations and the spatio-temporal information are not considered in constructing HLPF. To tackle their problems, we propose a set of novel high-level pose features (NHLPF). Specifically, considering that the distances between adjacent pairs of joints usually remain unchanged, we propose a horizontally relative position feature and a vertically relative position feature. In addition, a joint inner product feature is proposed to code the spatial information among each triplet of joints. To code temporal information, we calculate the trajectories of the above-mentioned three types of features as corresponding trajectory features. Furthermore, to combine the spatial and temporal information, we present a joint energy change feature, which is designed using observations of the magnitude and direction of the force between joints. We evaluate our NHLPF on a benchmark dataset. The results show that NHPLF are superior features for action recognition.

关键词： Trajectory Pose estimation Detectors Joining processes Hip Force Three-dimensional displays

来源：评论

学校读者我要写书评

暂无评论

Learned video compression with feature-level residuals

arXiv

引用

arXiv 2020年

作者： Feng, Runsen Wu, Yaojun Guo, Zongyu Zhang, Zhizheng Jin, Xin Chen, Zhibo CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

关键词： Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Hybrid digital-analog scheme for video transmission over fading channel

Hybrid digital-analog scheme for video transmission over fad...

引用

International Symposium on Circuits and systems

作者： Jian Shen Lei Yu Houqiang Li CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China

ISBN: (纸本)9781479953424

In this paper, we consider video communication over fading channel, where the perfect instantaneous channel state information (CSI) is available at both sender and receiver. Most of existing coding schemes are inefficient in this communication scenario. The reason is that for digital coding scheme, it has high coding efficiency but unavoidably leads to the cliff effect;while for analog scheme, it has graceful video quality variation with channel varying, but has low coding efficiency. Hence, to integrate the advantages of digital coding and analog coding, we propose a hybrid digital-analog (HDA) scheme. In our scheme, we have adopted adaptive power allocation and adaptive forward error coding (FEC) in digital part to accommodate instantaneous channel quality. The evaluation results show that the proposed HDA scheme outperforms ParCast (a state-of-the-art analog scheme) 0.3~2.2dB under the channel Signal-to-Noise Ratio (SNR) from 3dB to 20dB.

关键词： Encoding Resource management Signal to noise ratio Receivers Decoding Forward error correction Fading channels

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：