检索结果-内蒙古大学图书馆

Unsupervised Learning of 3D Scene Flow from Monocular Camera

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Guangming Tian, Xiaoyu Ding, Ruiqi Wang, Hesheng Department of Automation Institute of Medical Robotics Key Laboratory of System Control and Information Processing of Ministry of Education Key Laboratory of Marine Intelligent Equipment and System of Ministry of Education Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai 200240 China

Scene flow represents the motion of points in the 3D space, which is the counterpart of the optical flow that represents the motion of pixels in the 2D image. However, it is difficult to obtain the ground truth of scene flow in the real scenes, and recent studies are based on synthetic data for training. Therefore, how to train a scene flow network with unsupervised methods based on real-world data shows crucial significance. A novel unsupervised learning method for scene flow is proposed in this paper, which utilizes the images of two consecutive frames taken by monocular camera without the ground truth of scene flow for training. Our method realizes the goal that training scene flow network with real-world data, which bridges the gap between training data and test data and broadens the scope of available data for training. Unsupervised learning of scene flow in this paper mainly consists of two parts: (i) depth estimation and camera pose estimation, and (ii) scene flow estimation based on four different loss functions. Depth estimation and camera pose estimation obtain the depth maps and camera pose between two consecutive frames, which provide further information for the next scene flow estimation. After that, we used depth consistency loss, dynamic-static consistency loss, Chamfer loss, and Laplacian regularization loss to carry out unsupervised training of the scene flow network. To our knowledge, this is the first paper that realizes the unsupervised learning of 3D scene flow from monocular camera. The experiment results on KITTI show that our method for unsupervised learning of scene flow meets great performance compared to traditional methods Iterative Closest Point (ICP) and Fast Global Registration (FGR). The source code is available at: https://***/IRMVLab/3DUnMonoFlow. Copyright © 2022, The Authors. All rights reserved.

关键词： Unsupervised learning

Spatial public goods games on any population structure

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Chaoqian Su, Qi Department of Computational and Data Sciences George Mason University FairfaxVA22030 United States Department of Automation Shanghai Jiao Tong University Shanghai China Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai China

Understanding the emergence of cooperation in spatially structured populations has advanced significantly in the context of pairwise games, but the fundamental theory of group-based public goods games (PGGs) remains less explored. Here, we provide theoretical conditions under which cooperation thrive in spatial PGGs on any population structure, which are accurate under weak selection. We find that PGGs can support cooperation across all kinds of model details and on almost all network structures in contrast to pairwise games. For example, a class of networks that would otherwise fail to produce cooperation, such as star graphs, are particularly conducive to cooperation in spatial PGGs. This fundamental advantage of spatial PGGs derives from reciprocity through second-order interactions, allowing local structures such as the clustering coefficient to play positive roles. We also verify the robustness of spatial PGGs on empirical networks where pairwise games cannot support cooperation, which implies that PGGs could be a universal interaction mode in real-world systems. © 2024, CC BY.

关键词：

3D Instances as 1D Kernels

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wu, Yizheng Shi, Min Du, Shuaiyuan Lu, Hao Cao, Zhiguo Zhong, Weicai Key Laboratory of Image Processing and Intelligent Control Ministry of Education School of AIA Huazhong University of Science and Technology China Huawei CBG Consumer Cloud Service Search & Maps BU China

We introduce a 3D instance representation, termed instance kernels, where instances are represented by one-dimensional vectors that encode the semantic, positional, and shape information of 3D instances. We show that instance kernels enable easy mask inference by simply scanning kernels over the entire scenes, avoiding the heavy reliance on proposals or heuristic clustering algorithms in standard 3D instance segmentation pipelines. The idea of instance kernel is inspired by recent success of dynamic convolutions in 2D/3D instance segmentation. However, we find it non-trivial to represent 3D instances due to the disordered and unstructured nature of point cloud data, e.g., poor instance localization can significantly degrade instance representation. To remedy this, we construct a novel 3D instance encoding paradigm. First, potential instance centroids are localized as candidates. Then, a candidate merging scheme is devised to simultaneously aggregate duplicated candidates and collect context around the merged centroids to form the instance kernels. Once instance kernels are available, instance masks can be reconstructed via dynamic convolutions whose weights are conditioned on instance kernels. The whole pipeline is instantiated with a dynamic kernel network (DKNet). Results show that DKNet outperforms the state of the arts on both ScanNetV2 and S3DIS datasets with better instance localization. Code is available: https://***/W1zheng/DKNet. Copyright © 2022, The Authors. All rights reserved.

关键词： Pipelines

Unequal Error Protection Rateless Codes for Spatial Audio Transmission

学校读者我要写书评

暂无评论

Unequal Error Protection Rateless Codes for Spatial Audio Tr...

International Conference on Communication Technology (ICCT)

作者： Junpeng Yin Yusun Fu Haobo Huang Jinhui Tang Ningbo Artificial Intelligence Institute Shanghai Jiao Tong University Ningbo China School of Electronic Information and Electrical Engineering Shanghai Jiao Tong University Shanghai China School of Electronic Information and Electrical Engineering the Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai China

ISBN: (数字)9798350363760

ISBN: (纸本)9798350363777

Spatial audio is an important component of the metaverse. With the rapid increase of the spatial resolution and the number of objects, the data volume of the spatial side information has grown larger. However, the low bit rate transmission of spatial side information in the object-based audio coding methods leads to the low-quality reconstruction of spatial side information with limited bandwidth. To address this issue, this paper constructs a cascade design of spatial audio coding and unequal error protection (UEP) rateless codes. Furthermore, the further protection of specific spatial side information is proposed for different scenarios, which effectively addresses the limitation of traditional spatial audio coding in side information protection. For the situation where the code length K of different importance levels is different, we propose an adaptive degree distribution (ADD), adding K to the selection probability as an influencing factor, which has better performance than other degree distribution for short code length. ADD is used to replace the multiple degree distribution strategy of the traditional unequal error protection scheme, and the block error rate (BER) is reduced nearly 10 times compared with the traditional scheme.

关键词： Codes Error analysis Metaverse Distribution strategy Spatial audio Bit rate Symbols Communications technology Protection Spatial resolution

Vehicle Scheduling with Exact Minimum Fleet-Size

学校读者我要写书评

暂无评论

SSRN

SSRN 2022年

作者： Shen, Yindong Li, Yuanyuan School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China Key Laboratory of Image Processing and Intelligent Control Huazhong University of Science and Technology Ministry of Education China

Vehicle scheduling plays a profound role in public transit planning. Vehicle scheduling problem (VSP) addresses the task of assigning vehicles to cover timetabled trips to minimize the fleet-size as well as operating costs. The fixed cost of acquiring and maintaining vehicles is much more expensive than the operating cost. It is of great theoretical and practical significance to construct the vehicle schedules with the exact minimum fleet-size (EMF). The theorem that the EMF is equal to the maximum incompatible degree of the timetabled trips is first proposed and proved. Based on the theorem and time compatibility between any two trips, this paper theoretically examines vehicle scheduling with EMF. Specifically, construction procedures of schedules with EMF are provided for VSP in a specific case as well as the general case. Furthermore, sensitivity analysis of the EMF and the schedule with EMF are carried out. The EMF is possible to decrease by 1 if a trip is deleted or the start/end time of a trip is changed in the timetable. © 2022, The Authors. All rights reserved.

关键词： Sensitivity analysis

Impact signal detection and recovery based on the novel potential well model

学校读者我要写书评

暂无评论

Impact signal detection and recovery based on the novel pote...

第40届中国控制会议

作者： Shangbin Jiao Qing Zhang Shuang Lei Ding Xin Zhang Qing Wang Shaanxi Key Laboratory of Complex System Control and Intelligent Information Processing Xi’an University of Technology

To further improve the output effect of the stochastic resonance system driven by the impact signal,this paper uses a novel potential well model as the nonlinear model of the *** the α stable noise environment,the detection performance of the stochastic resonance system driven by the impact signal is analyzed when the characteristic coefficient is used as the measurement *** effect of system parameters on the output effect is studied,and the detection of multiple impact signal are *** view of the distortion of the output signal waveform of the stochastic resonance system,the cause of the waveform distortion is analyzed based on the principle of particle dynamics,and the signal recovery system is *** output signal of the stochastic resonance system is recovered by using the similarity as the output system measurement *** simulation results show that the stochastic resonance system based on the novel potential well model has a better detection effect,and can effectively detect the time information of the impact *** recovery system can realize the estimation of the time-domain waveform of the impact signal under the background of strong noise.

关键词： Novel potential well model stochastic resonance impact signal detection signal recovery

A Real-time Framework of Multi-Camera Vehicle Tracking System

学校读者我要写书评

暂无评论

A Real-time Framework of Multi-Camera Vehicle Tracking Syste...

2024 IEEE International Conference on Real-Time Computing and Robotics, RCAR 2024

作者： Zhuang, Hanyang Chen, Xingyuan Lan, Jianglin Wang, Ying Lu, Zhaoxuan Wang, Chunxiang Yang, Ming University of Michigan Shanghai Jiao Tong University Joint Institute Shanghai Jiao Tong University Shanghai200240 China University of Glasgow James Watt School of Engineering GlasgowG12 8QQ United Kingdom Shanghai Jiao Tong University School of Mechanical Engineering Shanghai200040 China Shanghai Intelligent and Connected Vehicle R and D Center Co. LTD Shanghai201400 China Shanghai Jiao Tong University Department of Automation Shanghai200240 China Ministry of Education of China Key Laboratory of System Control and Information Processing Shanghai200240 China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai200240 China

ISBN: (纸本)9798350372601

Multi-camera vehicle tracking (MCVT) system is a key technology to build an intelligent city and intelligent transportation system. The MCVT system utilizes roadside monitoring devices and computing platforms to achieve city-level road traffic perception at a finer scale by tracking each vehicle on the road to obtain the trajectory. The existing MCVT system mainly focuses on dealing with the algorithm modules separately, without studying the real-time performance and system architecture of the entire pipeline. However, in real-world applications, many of the tricks in improving evaluation metrics are not applicable since there is endless traffic flow running all the time. Therefore, this paper proposes a real-time framework of the MCVT system to streamline all the algorithm modules with proper data flow. Firstly, a multi-threaded multi-vehicle tracking (MVT) module is developed to detect vehicles in each camera. Re-identification is implemented during MVT to provide the feature for tracking in the same camera and across different cameras. Then, after the vehicle leaves the camera field of view, the feature of vehicle trajectory is calculated by averaging the feature at each frame. Meanwhile, a background thread runs continuously to match the incoming feature of vehicle trajectory with past features by conducting a topology searching method. In the end, the vehicle trajectories with well-matched features are considered to be the same vehicle. The entire pipeline has been validated on two different datasets, the precision is 0.64 in the CityFlow dataset which uses 6 cameras simultaneously. In the customized dataset, the same method can still reach real-time when using 16 cameras at the same time with smaller network modules. © 2024 IEEE.

关键词： Vehicle locating systems

Front-end Replication Dynamic Window (FRDW) for Online Motor imagery Classification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Chen, Xinru An, Jiayu Wu, Huanyu Li, Siyang Liu, Bin Wu, Dongrui The Ministry of Education Key Laboratory of Image Processing and Intelligent Control School of Artificial Intelligence and Automation Huazhong University of Science and Technology Wuhan430074 China The Shenzhen Huazhong University of Science and Technology Research Institute Shenzhen China Beijing Fenghuo Wanjia Technology Co. Ltd. Beijing100026 China

Motor imagery (MI) is a classical paradigm in electroencephalogram (EEG) based brain-computer interfaces (BCIs). Online accurate and fast decoding is very important to its successful applications. This paper proposes a simple yet effective front-end replication dynamic window (FRDW) algorithm for this purpose. Dynamic windows enable the classification based on a test EEG trial shorter than those used in training, improving the decision speed;front-end replication fills a short test EEG trial to the length used in training, improving the classification accuracy. Within-subject and cross-subject online MI classification experiments on three public datasets, with three different classifiers and three different data augmentation approaches, demonstrated that FRDW can significantly increase the information transfer rate in MI decoding. Additionally, FR can also be used in training data augmentation. FRDW helped win national champion of the China BCI Competition in 2022. © 2024, CC BY.

关键词： Brain computer interface

Fire Detection Based on Improved Feature Fusion

学校读者我要写书评

暂无评论

Fire Detection Based on Improved Feature Fusion

IEEE International Symposium on information (IT) in Medicine and Education, ITME

作者： Xinrong Cao Jincai Wu Zuoyong Li Fujian Provincial Key Laboratory of Information Processing and Intelligent Control College of Computer and Control Engineering Minjiang University Fuzhou China School of Electronic Electrical Engineering and Physics Fujian University of Technology Fuzhou China Sichuang Technology Co. Ltd. Fuzhou China

Fire is a serious disaster that endangers life and property worldwide. Accurate and prompt fire detection can provide early warning information for firefighting efforts. Deep learning-based object detection algorithms have proven to be useful in the field of fire detection. However, fire detection is a particularly challenging task, due to the low-resolution of fire images and the presence of blurring and deformation in the targets. Existing fire detection methods are limited by poor detection accuracy and slow speed in complex fire scenarios. To address these challenges, we propose a fire detection algorithm based on improved YOLOv5s. Firstly, the attention module is embedded into the network structure. We introduce this attention module into the last layer of the backbone to enhance feature extraction ability. Secondly, we modify the Neck structure to fully fuse different feature maps. Through fusing extra low-level features from the backbone, it utilizes various receptive fields to enhance the expression of feature. Thirdly, SIoU is introduced to achieve faster convergence during the training. SIoU take into account the direction of the mismatch to redefine penalty indicators. The experimental results show that our proposed algorithm outperforms commonly used models. Even compared to specific fire detectors, our model exhibits certain advantages.

关键词：