检索结果-内蒙古大学图书馆

Ransp: Ranking Attention Network For Saliency Prediction On Omnidirectional Images

学校读者我要写书评

暂无评论

Ransp: Ranking Attention Network For Saliency Prediction On ...

IEEE International Conference on Multimedia and Expo (ICME)

作者： Dandan Zhu Yongqing Chen Tian Han Defang Zhao Yucheng Zhu Qiangqiang Zhou Guangtao Zhai Xiaokang Yang Artificial Intelligence Institute Shanghai Jiao Tong University Hainan Air Traffic Management Sub-Bureau Department of Computer Science Stevens Institute of Technology School of Software Engineering Tongji University Institute of Image Communication and Network Engineering Shanghai Jiao Tong University School of Information and Computer Shanghai Business School

ISBN: (数字)9781728113319

ISBN: (纸本)9781728113326

Various convolutional neural network (CNN)-based methods have shown the ability to boost the performance of saliency prediction on omnidirectional images (ODIs). However, these methods are limited by sub-optimal accuracy, because not all the features extracted by the CNN model are not useful for the final fine-grained saliency prediction. Features are redundant and have negative impact on the final fine-grained saliency prediction. To tackle this problem, we propose a novel Ranking Attention Network for saliency prediction (RANSP) of head fixations on ODIs. Specifically, the part-guided attention (PA) module and channel-wise feature (CF) extraction module are integrated in a unified framework and are trained in an end-to-end manner for fine-grained saliency prediction. To better utilize the channel-wise feature map, we further propose a new Ranking Attention Module (RAM), which automatically ranks and selects these maps based on scores for fine-grained saliency prediction. Extensive experiments are conducted to show the effectiveness of our method for saliency prediction of ODIs.

关键词： Feature extraction Head Predictive models Random access memory Two dimensional displays Task analysis

Joint channel assignment and power allocation for multi-UAV communication

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhou, Lingyun Chen, Xihan Hong, Mingyi Jin, Shi Shi, Qingjiang School of Software Engineering Tongji University Shanghai200092 China Department of Information Science and Electronic Engineering Zhejiang University Hangzhou310000 China Department of Electrical and Computer Engineering University of Minnesota MinneapolisMN55455 United States National Mobile Communications Research Laboratory Southeast University Nanjing210096 China

Unmanned aerial vehicle (UAV) swarm has emerged as a promising novel paradigm to achieve better coverage and higher capacity for future wireless network by exploiting the more favorable line-of-sight (LoS) propagation. To reap the potential gains of UAV swarm, the remote control signal sent by ground control unit (GCU) is essential, whereas the control signal quality are susceptible in practice due to the effect of the adjacent channel interference (ACI) and the external interference (EI) from radiation sources distributed across the region. To tackle these challenges, this paper considers priority-aware resource coordination in a multi-UAV communication system, where multiple UAVs are controlled by a GCU to perform certain tasks with a pre-defined trajectory. Specifically, we maximize the minimum signal-to-interference-plus-noise ratio (SINR) among all the UAVs by jointly optimizing channel assignment and power allocation strategy under stringent resource availability constraints. According to the intensity of ACI, we consider the corresponding problem in two scenarios, i.e., Null-ACI and ACI systems. By virtue of the particular problem structure in Null-ACI case, we first recast the formulation into an equivalent yet more tractable form and obtain the global optimal solution via Hungarian algorithm. For general ACI systems, we develop an efficient iterative algorithm for its solution based on the smooth approximation and alternating optimization methods. Extensive simulation results demonstrate that the proposed algorithms can significantly enhance the minimum SINR among all the UAVs and adapt the allocation of communication resources to diverse mission priority. Copyright © 2020, The Authors. All rights reserved.

关键词： Unmanned aerial vehicles (UAV)

Modified indexing algorithm based on priority queue in metric space for MVP tree 4

学校读者我要写书评

暂无评论

Modified indexing algorithm based on priority queue in metri...

4th International Conference on Future Networks and Distributed Systems, ICFNDS 2020

作者： Fomin, Vladimir Aleksandrov, Igor Gallyamov, Denis Kirichek, Ruslan Dept. of Information Systems Herzen State Pedagogical University of Russia Saint Petersburg Russia Dept. of Software Engineering and Computer Science Bonch-Bruevich Saint Petersburg State University of Telecommunications Saint Petersburg Russia Bonch-Bruevich Saint Petersburg State University of Telecommunications Saint Petersburg Russia V. A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences Moscow Russia

ISBN: (纸本)9781450388863

The Internet of Things (IoT) algorithms process huge amounts of heterogeneous data in real-time. One of the most computationally intensive tasks using cloud technologies is the task of clustering and classifying data. The authors propose to develop an approach to data classification within the "Query by Similarity"paradigm, which uses the technology of data indexing based on Metric Access Methods (MAM). To improve the performance of data indexing, this paper proposes a similar nearest neighbor search method combining a multiple vantage point tree (MVP) and improved algorithms for processing the priority queue of nodes. The following two algorithms for processing the priority queue of nodes were developed: 1) algorithm for all kinds of points-queries, which makes it possible to take into account parent nodes of all higher levels;2) algorithm for grouped based on clustering of points-queries by reusing previously obtained search results. Experimental results confirm the effectiveness of the proposed approaches and algorithms. © 2020 ACM.

关键词： Nearest neighbor search

Generating superpixels for high-resolution images with Decoupled Patch Calibration

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Yaxiong Wei, Yunchao Qian, Xueming Zhu, Li Yang, Yi The School of Software Engineering Xi'an Jiaotong University Xi'an710049 China School of Computer and Information Technology Beijing Jiaotong University Beijing100000 China The Key Laboratory for Intelligent Networks and Network Security Ministry of Education Xi'an Jiaotong University Xi'an710049 China The SMILES Laboratory Xi'an Jiaotong University Xi'an710049 China Zhibian Technology Co. Ltd. Taizhou317000 China The School of Software Xi'an Jiaotong University Xi'an710049 China School of Computer Science and Technology Zhejiang University Hangzhou310000 China

Superpixel segmentation has recently seen important progress benefiting from the advances in differentiable deep learning. However, the very high-resolution superpixel segmentation still remains challenging due to the expensive memory and computation cost, making the current advanced superpixel networks fail to process. In this paper, we devise Patch Calibration Networks (PCNet), aiming to efficiently and accurately implement high-resolution superpixel segmentation. PCNet follows the principle of producing high-resolution output from low-resolution input for saving GPU memory and relieving computation cost. To recall the fine details destroyed by the down-sampling operation, we propose a novel Decoupled Patch Calibration (DPC) branch for collaboratively augment the main superpixel generation branch. In particular, DPC takes a local patch from the high-resolution images and dynamically generates a binary mask to impose the network to focus on region boundaries. By sharing the parameters of DPC and main branches, the fine-detailed knowledge learned from high-resolution patches will be transferred to help calibrate the destroyed information. To the best of our knowledge, we make the first attempt to consider the deep-learning-based superpixel generation for high-resolution cases. To facilitate this research, we build evaluation benchmarks from two public datasets and one new constructed one, covering a wide range of diversities from fine-grained human parts to cityscapes. Extensive experiments demonstrate that our PCNet can not only perform favorably against the state-of-the-arts in the quantitative results but also improve the resolution upper bound from 3K to 5K on 1080Ti GPUs. © 2021, CC BY.

关键词： Image segmentation

Direct Image to point cloud descriptors matching for 6-dof camera localization in dense 3d point Cloud

学校读者我要写书评

暂无评论

arXiv 2019年

作者： Nadeem, Uzair Jalwana, Mohammad A. A. K. Bennamoun, Mohammed Togneri, Roberto Sohel, Ferdous Department of Computer Science and Software Engineering University of Western Department of Electrical Electronics and Computer Engineering University of Western School of Engineering and Information Technology Murdoch University

We propose a novel concept to directly match feature descriptors extracted from RGB images, with feature descriptors extracted from 3D point clouds. We use this concept to localize the position and orientation (pose) of the camera of a query image in dense point clouds. We generate a dataset of matching 2D and 3D descriptors, and use it to train a proposed Descriptor-Matcher algorithm. To localize a query image in a point cloud, we extract 2D keypoints and descriptors from the query image. Then the Descriptor- Matcher is used to find the corresponding pairs 2D and 3D keypoints by matching the 2D descriptors with the preextracted 3D descriptors of the point cloud. This information is used in a robust pose estimation algorithm to localize the query image in the 3D point cloud. Experiments demonstrate that directly matching 2D and 3D descriptors is not only a viable idea but also achieves competitive accuracy compared to other state-of-the-art approaches for camera pose localization. Copyright © 2019, The Authors. All rights reserved.

关键词： Cameras

The computer Vision-based Tolerancing Callout Detection Model

学校读者我要写书评

暂无评论

Procedia CIRP 2020年 92卷 134-139页

作者： Xin Deng Tukun Li Yuanping Xu Yanlong Cao Chao Kong Errui Zhang School of Software Engineering Chengdu University of Information Technology Chengdu China EPSRC Future Advanced Metrology Hub Centre for Precision Technologies School of Computing and Engineering University of Huddersfield Huddersfield HD1 3DH UK Shandong Industrial Technology Research Institute of Zhejiang University Shandong 277500 China College of Engineering and Computer Science Australia National University Canberra ACT 2600 Australia

Tolerancing symbols play an important role in mechanical product drawings, and they directly determine the functions, mating properties, interchangeability and working life of geometrical products. A symbolic tolerancing callout containing a set of symbols represents a set of pre-ordered operations with attributes. It includes an amount of knowledge from the drawing and standard documents, which is often reconstructed manually by engineers. Thus, at the same time, a symbolic tolerancing callout makes it difficult for the end-user to understand and interpret these callouts manually. To this end, this study puts forward a tolerancing callout detection model via the use of off-the-shelf costumer-grade cameras on current mobile devices for extracting and recognizing tolerancing callout blocks and symbols in them intelligently. This model has four core components: image preprocessing, callout location and extraction, symbol and character segmentation, and deep learning-based symbol recognition. The image preprocessing component is developed to remove the interferences on the target technical drawings through the corresponding morphological methods. This study proposes a novel solution on callout block locations and extractions in callout intensive scenarios since the callout locations and extractions can directly affect the accuracy of symbol and character recognitions. Then, Huff Transform and improved projection methods have been devised to symbol and character segmentations. Finally, this study constructed a convolutional neural network (CNN) to train a symbol recognition model. The experimental results show that the proposed model gains applicability on intelligent callout extractions and the corresponding symbol recognitions.

关键词： Tolerancing symbols Callout location extraction Symbol character segmentation CNN

A Review of Artificial Fish Swarm Algorithms: Recent Advances and Applications

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Pourpanah, Farhad Wang, Ran Lim, Chee Peng Wang, Xi-Zhao Yazdani, Danial College of Mathematics and Statistics Guangdong Key Lab. of Intelligent Information Processing Shenzhen University China Department of Electrical and Computer Engineering University of Windsor Canada College of Mathematics and Statistics Shenzhen Key Lab. of Advanced Machine Learning and Applications Guangdong Key Lab. of Intelligent Information Processing Shenzhen University China Institute for Intelligent Systems Research and Innovation Deakin University Australia College of Computer Science and Software Engineering Guangdong Key Lab. of Intelligent Information Processing Shenzhen University China School of Computer Science and Engineering Southern University of Science and Technology China

The Artificial Fish Swarm Algorithm (AFSA) is inspired by the ecological behaviors of fish schooling in nature, viz., the preying, swarming and following behaviors. Owing to a number of salient properties, which include flexibility, fast convergence, and insensitivity to the initial parameter settings, the family of AFSA has emerged as an effective Swarm Intelligence (SI) methodology that has been widely applied to solve real-world optimization problems. Since its introduction in 2002, many improved and hybrid AFSA models have been developed to tackle continuous, binary, and combinatorial optimization problems. This paper aims to present a concise review of the continuous AFSA, encompassing the original ASFA, its improvements and hybrid models, as well as their associated applications. We focus on articles published in high-quality journals since 2013. Our review provides insights into AFSA parameters modifications, procedure and sub-functions. The main reasons for these enhancements and the comparison results with other hybrid methods are discussed. In addition, hybrid, multi-objective and dynamic AFSA models that have been proposed to solve continuous optimization problems are elucidated. We also analyse possible AFSA enhancements and highlight future research directions for advancing AFSA-based models. Copyright © 2020, The Authors. All rights reserved.

关键词： Multiobjective optimization

Analyzing Data Mining and Its Application to Smart Business

学校读者我要写书评

暂无评论

Analyzing Data Mining and Its Application to Smart Business

Technology Innovation Management and engineering science International Conference (TIMES-iCON)

作者： Saifun Nahar Ting Zhong Happy N. Monday Grace U. Nneji Michael O. Mills Hassan S. Abubakar School of Information and Software Engineering School of Computer Science and Engineering School of Information and Communication Engineering University of Electronic Science and Technology of China Chengdu Sichuan P.R China

ISBN: (数字)9781728137551

ISBN: (纸本)9781728137568

This paper presents the main concepts of smart business (SB) and the application of smart business to provide solutions to the problems faced by a corporation. In this context, the main emphasis is on the approach of data mining including one of the tools used by SB that uses Decision Trees and Artificial Intelligence techniques along with other techniques. Moreover, algorithms such as Genetic Algorithms system and the Neural Networks are considered in its implementation. The concept of Smart Business (SB) has clarified these problems and consists of broad category technologies and application programs used to extract, store, analyze and transform high dimensional data. The approach of Smart Business is presented according to the concepts of Data Warehouse, Data Mart and Data Mining.

关键词： Data mining Tools Data warehouses Companies Decision trees Neural networks

A Survey on Data Stream Mining Towards the Internet of Things Application

学校读者我要写书评

暂无评论

A Survey on Data Stream Mining Towards the Internet of Thing...

Technology Innovation Management and engineering science International Conference (TIMES-iCON)

作者： Saifun Nahar Ting Zhong Happy N. Monday Michael O. Mills Grace U. Nneji Hassan S. Abubakar School of Information and Software Engineering School of Computer Science and Engineering School of Information and Communication Engineering University of Electronic Science and Technology of China ChengduSichuan P.R China

ISBN: (数字)9781728137551

ISBN: (纸本)9781728137568

In the era of the Internet of Things, a widespread of applications depend on time with the various number of different data generated and collected from different devices available. These devices depend on the type of application. These fast stream data are real-time and large in dimension for the purpose of making decision as well as predicting future occurrence and analytics. Datastream analytics of internet technology for both businesses and everyday life is very valuable in terms of developing good quality of life. In this study, first of all, we focus on the concept of Internet of Things and its relationship with its architecture, large and flowing data. In addition, the approach of Internet of Things applied knowledge discovery process and deep learning frameworks are presented in this paper. Finally, the Internet of Things and its features are introduced in this work as well as the commonly used tools.

关键词： Internet of Things Machine learning Big Data Data mining Data analysis Sensors Python