检索结果-内蒙古大学图书馆

Camouflaged Object Detection via Context-aware Cross-level Fusion

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Chen, Geng Liu, Si-Jie Sun, Yu-Jia Ji, Ge-Peng Wu, Ya-Feng Zhou, Tao Northwestern Polytechnical University Xi’an710072 China National Engineering Laboratory for Integrated Aero-Space-Ground-Ocean Big Data Application Technology School of Computer Science and Engineering China Data Processing Center School of Power and Energy China Key Laboratory of System Control and Information Processing Ministry of Education Shanghai200240 China School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing210094 China School of Computer Science Inner Mongolia University Hohhot010021 China Artificial Intelligence Institute School of Computer Science Wuhan University Wuhan430072 China

Camouflaged object detection (COD) aims to identify the objects that conceal themselves in natural scenes. Accurate COD suffers from a number of challenges associated with low boundary contrast and the large variation of object appearances, e.g., object size and shape. To address these challenges, we propose a novel Context-aware Cross-level Fusion Network (C2F-Net), which fuses context-aware cross-level features for accurately identifying camouflaged objects. Specifically, we compute informative attention coefficients from multi-level features with our Attention-induced Cross-level Fusion Module (ACFM), which further integrates the features under the guidance of attention coefficients. We then propose a Dual-branch Global Context Module (DGCM) to refine the fused features for informative feature representations by exploiting rich global context information. Multiple ACFMs and DGCMs are integrated in a cascaded manner for generating a coarse prediction from high-level features. The coarse prediction acts as an attention map to refine the low-level features before passing them to our Camouflage Inference Module (CIM) to generate the final prediction. We perform extensive experiments on three widely used benchmark datasets and compare C2F-Net with state-of-the-art (SOTA) models. The results show that C2F-Net is an effective COD model and outperforms SOTA models remarkably. Further, an evaluation on polyp segmentation datasets demonstrates the promising potentials of our C2F-Net in COD downstream applications. Our code is publicly available at: https://***/Ben57882/C2FNet-TSCVT. Copyright © 2022, The Authors. All rights reserved.

关键词： Object detection

Learned Image Compression with Gaussian-Laplacian-Logistic Mixture Model and Concatenated Residual Modules

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Fu, Haisheng Liang, Feng Lin, Jianping Li, Bing Akbari, Mohammad Liang, Jie Zhang, Guohe Liu, Dong Tu, Chengjie Han, Jingning The School of Microelectronics Xi’an Jiaotong University Xi’an China The CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China The School of Engineering Science Simon Fraser University Canada The Tencent Technologies China The Google Inc. United States

Recently deep learning-based image compression methods have achieved significant achievements and gradually outperformed traditional approaches including the latest standard Versatile Video Coding (VVC) in both PSNR and MS-SSIM metrics. Two key components of learned image compression are the entropy model of the latent representations and the encoding/decoding network architectures. Various models have been proposed, such as autoregressive, softmax, logistic mixture, Gaussian mixture, and Laplacian. Existing schemes only use one of these models. However, due to the vast diversity of images, it is not optimal to use one model for all images, even different regions within one image. In this paper, we propose a more flexible discretized Gaussian-Laplacian-Logistic mixture model (GLLMM) for the latent representations, which can adapt to different contents in different images and different regions of one image more accurately and efficiently, given the same complexity. Besides, in the encoding/decoding network design part, we propose a concatenated residual blocks (CRB), where multiple residual blocks are serially connected with additional shortcut connections. The CRB can improve the learning ability of the network, which can further improve the compression performance. Experimental results using the Kodak, Tecnick-100 and Tecnick-40 datasets show that the proposed scheme outperforms all the leading learning-based methods and existing compression standards including VVC intra coding (4:4:4 and 4:2:0) in terms of the PSNR and MS-SSIM. The source code is available at https://***/fengyurenpingsheng. Copyright © 2021, The Authors. All rights reserved.

关键词： Image compression

SAR image classification based on spiking neural network through spike-time dependent plasticity and gradient descent

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Chen, Jiankun Qiu, Xiaolan Ding, Chibiao Wu, Yirong National Key Laboratory of Science and Technology on Microwave Imaging Institute of Electronics Aerospace Information Research Institute Beijing10090 China The Key Laboratory of Technology in Geo-spatial Information Processing and Application System Aerospace Information Research Institute Chinese Academy of Sciences Beijing10090 China Aerospace Information Research Institute Chinese Academy of Sciences Beijing100190 China Suzhou Research Institute Aerospace Information Research Institute Chinese Academy of Sciences Suzhou215123 China The University of Chinese Academy of Sciences Beijing10049 China

At present, the Synthetic Aperture Radar (SAR) image classification method based on convolution neural network (CNN) has faced some problems such as poor noise resistance and generalization ability. Spiking neural network (SNN) is one of the core components of brain-like intelligence and has good application prospects. This article constructs a complete SAR image classifier based on unsupervised and supervised learning of SNN by using spike sequences with complex spatio-temporal information. We firstly expound the spiking neuron model, the receptive field of SNN, and the construction of spike sequence. Then we put forward an unsupervised learning algorithm based on STDP and a supervised learning algorithm based on gradient descent. The average classification accuracy of single layer and bilayer unsupervised learning SNN in three categories images on MSTAR dataset is 80.8% and 85.1%, respectively. Furthermore, the convergent output spike sequences of unsupervised learning can be used as teaching signals. Based on the TensorFlow framework, a single layer supervised learning SNN is built from the bottom, and the classification accuracy reaches 90.05%. By comparing noise resistance and model parameters between SNNs and CNNs, the effectiveness and outstanding advantages of SNN are verified. Code to reproduce our experiments is available at https://***/Jiankun-chen/Supervised-SNN-with-GD. © 2021, CC BY.

关键词： Gradient methods

Analysis of Landscape Pattern Spatial Scale in Middle and Upper Reaches of Meijiang River Basin 7th

学校读者我要写书评

暂无评论

Analysis of Landscape Pattern Spatial Scale in Middle and Up...

7th International Conference on Geoinformatics in Sustainable Ecosystem and Society, GSES 2019, and 1st International Conference on geospatial Artificial Intelligence for Urban Computing, GeoAI 2019

作者： Chen, Yuchan Zhang, Zhengdong Yang, Chuanxun Yang, Yang Zhang, Chen Han, Liusheng Yang, Ji Han, Xiangyu Key Lab of Guangdong for Utilization of Remote Sensing and Geographical Information System Guangdong Open Laboratory of Geospatial Information Technology and Application Guangzhou Institute of Geography Guangzhou510070 China School of Geography South China Normal University Guangzhou510631 China Guangzhou511458 China University of Chinese Academy of Sciences Beijing100049 China State Grid Chengdu Electric Supply Company Chengdu610000 China

ISBN: (纸本)9789811561054

The scale research of landscape pattern is an important basis for the study of spatiotemporal evolution of landscape pattern and the scientific and reasonable allocation of landscape pattern. This paper takes the middle and upper reaches of Meijiang river as the research area. Combined with spatial grain size analysis and extent size analysis, the spatial scale effect was studied to select the optimal research scale in this research area. The results show that most of the appropriate grain size of landscape pattern indexes are in range of 90–150 m, and the optimal grain size scale is 150 m in Middle and Upper Reaches of Meijiang River Basin. At the grain size scale of 150 m, the spatial self-correlation of landscape pattern index over 50% is the highest at the extent size of 300 m, and the optimal extent size scale is 300 m in this basin. This paper determines the optimal scale of landscape pattern in the middle and upper reaches of Meijiang river basin, which is of great significance to the ecological balance and sustainable development of the basin. © 2020, Springer Nature Singapore Pte Ltd.

关键词： Rivers

Stereo Matching of VHR Remote Sensing Images via Bidirectional Pyramid Network

学校读者我要写书评

暂无评论

Stereo Matching of VHR Remote Sensing Images via Bidirection...

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Rongshu Tao Yuming Xiang Hongjian You Key Laboratory of Technology in Geo-spatial Information Processing and Application System Aerospace Information Research Institute Chinese Academy of Science Beijing China University of Chinese Academy of Sciences Beijing China

ISBN: (数字)9781728163741

ISBN: (纸本)9781728163758

Stereo matching, an essential step in 3D reconstruction, still faces unignorable problems due to the very high resolution and complex structures of remote sensing images. Especially in occluded areas of high buildings and untextured areas of waters and woods, precise disparity estimation has become a difficult but important task. In this paper, we propose a novel method based on the pyramid stereo matching network to solve the aforementioned problems. Inspired by the classical optical flow estimation framework, we adopt the forward-backward consistency assumption to improve the accuracy. Moreover, we improve the construction of cost volume since the traditional deep-learning networks only work well for positive disparities and the disparity ranges in remote sensing images vary a lot. The proposed network is compared with two baselines. The experimental results show that our proposed method outperforms two baselines in terms of average endpoint error (EPE) and the fraction of erroneous pixels(D1), and the improvements in occluded areas are significant.

关键词： Estimation Remote sensing Optical sensors Optical imaging Three-dimensional displays Buildings Satellites

Sequential reinforced 360-degree video adaptive streaming with cross-user attentive network

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Fu, Jun Chen, Zhibo Chen, Xiaoming Li, Weiping CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China School of Computer Science and Engineering Beijing Technology and Business University China

In the tile-based 360-degree video streaming, predicting user’s future viewpoints and developing adaptive bitrate (ABR) algorithms are essential for optimizing user’s quality of experience (QoE). Traditional single-user based viewpoint prediction methods fail to achieve good performance in long-term prediction, and the recently proposed reinforcement learning (RL) based ABR schemes applied in traditional video streaming can not be directly applied in the tile-based 360-degree video streaming due to the exponential action space. Therefore, we propose a sequential reinforced 360-degree video streaming scheme with cross-user attentive network. Firstly, considering different users may have the similar viewing preference on the same video, we propose a cross-user attentive network (CUAN), boosting the performance of long-term viewpoint prediction by selectively utilizing cross-user information. Secondly, we propose a sequential RL-based (360SRL) ABR approach, transforming action space size of each decision step from exponential to linear via introducing a sequential decision structure. We evaluate the proposed CUAN and 360SRL using trace-driven experiments and experimental results demonstrate that CUAN and 360SRL outperform existing viewpoint prediction and ABR approaches with a noticeable margin. Copyright © 2020, The Authors. All rights reserved.

关键词： Forecasting

Phase Transition Analysis for Compressive Sensing Based DoA Estimation in Automotive Radar

学校读者我要写书评

暂无评论

Phase Transition Analysis for Compressive Sensing Based DoA ...

International Radar Symposium (IRS)

作者： Yan Zhang Silin Gao Mingqian Liu Bingchen Zhang Yirong Wu Chinese Academy of Sciences Aerospace Information Research Institute Beijing China University of Chinese Academy of Sciences Beijing China Key Laboratory of Technology in Geo-spatial Information and Processing Application System Beijing China

ISBN: (数字)9788394942151

ISBN: (纸本)9781728157863

In automotive radar applications, the compressive sensing (CS) based DoA estimation is used in array signal processing in recent years. Sparse reconstruction has the potential to estimate the direction of arrival (DoA) with super resolution. However, failed results may be acquired via sparse reconstruction in inappropriate conditions, namely the critical condition determining success or failure must be taken into consideration. In this paper, the sparsity of the scenario and the signal-to-noise ratio (SNR) are analyzed as the main factors via phase transition diagrams. Other factors affecting the success or failure are also investigated, such as the array configuration and the sparse recovery algorithm. Simulated and experimental results demonstrate the critical conditions, in which the DoA estimation is successful or failed.

关键词： Direction-of-arrival estimation Estimation Radar applications Compressed sensing Signal resolution Signal to noise ratio Automotive engineering

On the Use of Axial Ratio to Analysis Transmit Distortions in Hybrid Polarimetry and Their Influence to Canonical Scattering Mechanisms 6

学校读者我要写书评

暂无评论

On the Use of Axial Ratio to Analysis Transmit Distortions i...

6th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2019

作者： Chen, Shiqiang Hong, Wen Key Laboratory of Technology in Geo-spatial Information Processing and Application System Beijing China Institute of Electronics Chinese Academy of Sciences Beijing China University of Chinese Academy of Sciences Beijing China

ISBN: (纸本)9781728129129

Transmit distortions in the hybrid polarimetric (HP) SAR cannot be compensated simply with external calibration methods. Therefore, it is necessary to analysis their influence on HP data. In this study, we have proposed an analysis method based on the Axial Ratio (AR) of the real transmitted polarized wave. Firstly, impact of different distortion sources is quantified with the AR parameter. Next, a case study is illustrated using real GF-3 measured data. At last, the influence of transmit distortions to canonical scattering mechanisms is analyzed. © 2019 IEEE.

关键词： Polarimeters

Prior-enlightened and motion-robust video deblurring

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhou, Ya Xu, Jianfeng Tasaka, Kazuyuki Chena, Zhibo Li, Weiping CAS Key Laboratory of Technology in Geo-Spatial Information Processing and Application System University of Science and Technology of China Hefei230026 China KDDI Research Inc. Fujimino-shi Saitama356-8502 Japan

Various blur distortions in video will cause negative impact on both human viewing and video-based applications, which makes motion-robust deblurring methods urgently needed. Most existing works have strong dataset dependency and limited generalization ability in handling challenging scenarios, like blur in low contrast or severe motion areas, and non-uniform blur. Therefore, we propose a PRiOr-enlightened and MOTION-robust video deblurring model (PROMOTION) suitable for challenging blurs. On the one hand, we use 3D group convolution to efficiently encode heterogeneous prior information, explicitly enhancing the scenes’ perception while mitigating the output’s artifacts. On the other hand, we design the priors representing blur distribution, to better handle non-uniform blur in spatio-temporal domain. Besides the classical camera shake caused global blurry, we also prove the generalization for the downstream task suffering from local blur. Extensive experiments demonstrate we can achieve the state-of-the-art performance on well-known REDS and GoPro datasets, and bring machine task gain. Copyright © 2020, The Authors. All rights reserved.

关键词： Motion estimation