检索结果-内蒙古大学图书馆

3d multi-object tracking based on parallel multimodal data association

MACHINE VISION ANd APPLICATIONS 2025年第3期36卷 1-15页

作者： Tan, Shiyu Li, Xu Xu, Qimin Zhu, Jianxiao Southeast Univ Nanjing 210000 Jiangsu Peoples R China

This paper introduces a novel 3d multi-object tracker based on tracking-by-detection (TBd) framework. The system leverages parallel feature extraction and data association methods to process 2d appearance features and 3d spatial information, respectively, to achieve accurate tracking of targets in autonomous vehicles and intelligent transportation systems. By combining the Siamese network-based image feature extractor to extract image appearance features and the kinematic model established using Kalman filtering, our tracker effectively utilizes the appearance and spatial information of the object, thus improving tracker accuracy and reliability. Experimental results demonstrate the competitiveness of our tracker on the KITTI tracking benchmark. Compared to previous methods, A parallel feature extraction algorithm is proposed that can independently extract the spatial and appearance information of the object. A novel data association algorithm is designed that makes full use of the spatial information of the object from the point cloud and the appearance information from the image. This work provides substantial technical underpinnings for the advancement of autonomous driving and intelligent transportation technology. https://***/TanShiyu2022/PMTrack.

关键词： 3d multi-object tracking Camera and lidar fusion data association object detection and tracking

来源：评论

学校读者我要写书评

暂无评论

3d multi-object tracking based on informatic divergence-guided data association

引用

SIGNAL PROCESSING 2024年 222卷

作者： He, Jiawei Fu, Chunyun Wang, Xiyang Wang, Jianwen Chongqing Univ Coll Mech & Vehicle Engn Chongqing 400044 Peoples R China MEGVII Technol Beijing 100096 Peoples R China Chongqing Changan Automobile Co Ltd Chongqing 400023 Peoples R China

In the existing literature, most 3d multi-object tracking algorithms based on the tracking-by-detection framework employed deterministic tracks and detections for similarity calculation in the data association stage. Namely, the inherent uncertainties existing in tracks and detections are overlooked. In this work, we discard the commonly used deterministic tracks and deterministic detections for data association, instead, we propose to model tracks and detections as random vectors in which uncertainties are taken into account. Then, based on a modified Kullback-Leibler divergence, the similarity between two multidimensional distributions, i.e. track and detection, is evaluated for data association purposes. Lastly, the level of track uncertainty is incorporated in our cost function design to guide the data association process. Comparative experiments have been conducted on two typical datasets, KITTI and nuScenes, and the results indicated that our proposed method outperformed the compared state-of-the-art 3d tracking algorithms. For the benefit of the community, our code has been made available at https://***/hejiawei2023/UG3dMOT.

关键词： 3d multi-object tracking data association Cost function

来源：评论

学校读者我要写书评

暂无评论

3d multi-object tracking Based on dual-Tracker and d-S Evidence Theory

IEEE TRANSACTIONS ON INTELLIGENT VEHICLES

引用

IEEE TRANSACTIONS ON INTELLIGENT VEHICLES 2023年第3期8卷 2426-2436页

作者： Ma, Yuanzhi Zhang, Jindong Qin, Guihe Jin, Jingyi Zhang, Kunpeng Pan, dongyu Chen, Mai Jilin Univ Coll Comp Sci & Technol Changchun 130012 Peoples R China Minist Educ Key Lab Symbol Comp & Knowledge Engn Changchun 130012 Peoples R China State Key Lab Automobile Simulat & Control Changchun 130025 Peoples R China

Most of the current self-driving cars are equipped with a variety of sensors (such as lidar, camera), and it has become a trend to integrate multi-sensor information to achieve 3d multi-object tracking (MOT). In addition, unlike the perfect experimental conditions on public datasets, cars on real roads may experience a 3d sensor failure due to weather, mutual interference, etc. In this paper, we propose a dual-tracker-based 3d MOT system, which fuses 2d and 3d information to track objects and can still ensure good tracking accuracy when the 3d sensor fails in a single frame or multiple consecutive frames. Among them, two internal trackers complete the data association of 3d and 2d information, respectively. However, the outputs of the two internal trackers may conflict. First, we calculate the degree of association of potential matched pairs from the features extracted by each sensor, and then normalize it into the mass function of d-S evidence theory. Finally, we combine the evidence to obtain the final matched pairs of detection instances and track estimates. We do extensive experiments on the KITTI MOT dataset and simulate sensor failure scenarios by ignoring the output of the 3d detector. Using the latest evaluation measure for comparison, the results show that our method outperforms other advanced open-source 3d MOTs in a variety of scenarios.

关键词： Three-dimensional displays Sensors detectors Automobiles Evidence theory Trajectory Image edge detection 3d multi-object tracking d-S evidence theory multi-sensor fusion

来源：评论

学校读者我要写书评

暂无评论

3d multi-object tracking With Adaptive Cubature Kalman Filter for Autonomous driving

IEEE TRANSACTIONS ON INTELLIGENT VEHICLES

引用

IEEE TRANSACTIONS ON INTELLIGENT VEHICLES 2023年第1期8卷 512-519页

作者： Guo, Ge Zhao, Shijie Northeastern Univ State Key Lab Synthet Automat Proc Ind Shenyang 110819 Peoples R China Northeastern Univ Coll Informat Sci & Engn Shenyang 110819 Peoples R China

A crucial and challenging issue in autonomous driving is dynamic road environment detection and 3d multi object tracking. In this article, we propose a novel framework for online 3d multi-object tracking to eliminate the influence of inherent uncertainty and unknown biases in point cloud. A constant turn rate and velocity (CTRV) motion model is employed to estimate the future motion state, which are smoothed by a cubature Kalman filter (CKF) algorithm. A new affinity model is introduced to evaluate the similarity between trajectories and candidate detections for accurate and reliable data association which can be formulated as a bipartite matching problem. An adaptive cubature Kalman filter (ACKF) is given to remove the influence of unknown bias and to robustly update the tracked state. Accuracy and speed of the proposed tracking method are evaluated on the KITTI 3d multi-object tracking dataset, showing superior performance than the baselines.

关键词： tracking Three-dimensional displays Solid modeling Point cloud compression Predictive models Kalman filters Prediction algorithms 3d multi-object tracking point cloud data association Kalman filter (KF)

来源：评论

学校读者我要写书评

暂无评论

3d multi-object tracking in Autonomous driving: A survey 36

3D Multi-object Tracking in Autonomous Driving: A survey

引用

36th Chinese Control and decision Conference (CCdC)

作者： Peng, Zhe Liu, Wei Ning, Zuotao Zhao, Qixi Cheng, Shuai Hu, Jun Northeastern Univ Coll Informat Sci & Engn Shenyang Peoples R China Automot Technol Ltd Neusoft Reach Shenyang Peoples R China

ISBN: (纸本)9798350387780;9798350387797

3d multi-object tracking(3d MOT) is an indispensable component of autonomous driving because of its ability to perceive and track surrounding objects. In order to track the same object in consecutive frames, 3d MOT faces challenges such as object occlusion, abrupt motion, lighting variations and distortions, as well as scenarios with small and dense objects. In recent years, various methods have been proposed to address these issues. Therefore, it is crucial to analyze the commonalities and differences among different methods. Based on the relationship between detection and data association, this paper analyzes some mainstream methods in the recent three years. It categorizes them into three classes: tracking by detection(TBd), joint detection and embedding(JdE), and joint detection and tracking(JdT). Furthermore, for each method, we further divide it into several subcategories and summarize their approaches. Additionally, we compare the advantages and disadvantages of different methods and conduct a performance analysis on the nuScenes dataset.

关键词： 3d multi-object tracking Autonomous driving deep Learning

来源：评论

学校读者我要写书评

暂无评论

3d multi-object tracking in Point Clouds Based on Prediction Confidence-Guided data Association

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2022年第6期23卷 5668-5677页

作者： Wu, Hai Han, Wenkai Wen, Chenglu Li, Xin Wang, Cheng Xiamen Univ Fujian Key Lab Sensing & Comp Smart Cities Xiamen 361005 Peoples R China Xiamen Univ Sch Informat Xiamen 361005 Peoples R China Louisiana State Univ Sch Elect Engn & Comp Sci Baton Rouge LA 70803 USA

This paper proposes a new 3d multi-object tracker to more robustly track objects that are temporarily missed by detectors. Our tracker can better leverage object features for 3d multi-object tracking (MOT) in point clouds. The proposed tracker is based on a novel data association scheme guided by prediction confidence, and it consists of two key parts. First, we design a new predictor that employs a constant acceleration (CA) motion model to estimate future positions, and outputs a prediction confidence to guide data association through increased awareness of detection quality. Second, we introduce a new aggregated pairwise cost to exploit features of objects in point clouds for faster and more accurate data association. The proposed cost consists of geometry, appearance and motion components. Specifically, we formulate the geometry cost using resolutions (lengths, widths and heights), centroids, and orientations of 3d bounding boxes (BBs), the appearance cost using appearance features from the deep learning-based detector backbone network, and the motion cost by associating different motion vectors. Extensive multi-object tracking experiments on the KITTI tracking benchmark demonstrated that our method outperforms, by a large margin, the state-of-the-art methods in both tracking accuracy and speed.

关键词： Three-dimensional displays tracking Feature extraction detectors Two dimensional displays Predictive models Acceleration 3d multi-object tracking point clouds data association object detection and tracking

来源：评论

学校读者我要写书评

暂无评论

Co-MOT: Exploring the Collaborative Relations in Traffic Flow for 3d multi-object tracking

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2025年第4期26卷 4744-4756页

作者： Liu, Ye Liu, Xingdi Jiang, Zhongbin Liu, Jun Nanjing Univ Posts & Telecommun Coll Automation & Artificial Intelligence Nanjing 210023 Peoples R China Singapore Univ Technol & Design Informat Syst Technol & Design Pillar Singapore 487372 Singapore

In real-world scenes, vehicles and pedestrians on the road often exhibit consistency in their overall motion, forming the traffic flow we observe. Exploring this global collective motion consistency to aid in 3d multi-object tracking (MOT) tasks is an under-investigated issue in existing research. Recently, Graph Neural Networks (GNN) have been introduced to model interactions between targets in 3d tracking problems, achieving remarkable performance. However, existing GNN based methods usually employ neighborhood-based approaches to construct graphs which are unable to fully exploit collective relations in traffic flow. In this paper, we propose a GNN based 3d MOT method which effectively utilizes the collective motion consistency in traffic flow. Collective motion is modeled with a densely connected intra-flow graph within the collective group, allowing information to flow quickly. To build the intra-flow graph, we propose an effective collinearity condition to distinguish potential collective groups from the detected objects. For reasoning on the graph, we propose a progressive serial message-passing solver which enables the network to learn complex group movement relationships based on a thorough understanding of simple neighborhood relations. Our proposed method achieves state-of-the-art performance on public datasets: NuScenes and KITTI tracking benchmark. We have conducted extensive experiments to evaluate the comprehensive performance of our proposed method which demonstrates the effectiveness of the proposed method.

关键词： Three-dimensional displays tracking Graph neural networks Target tracking Solid modeling Feature extraction Collaboration Message passing Predictive models Prediction algorithms 3d multi-object tracking graph neural network collaborative relations collective motion

来源：评论

学校读者我要写书评

暂无评论

MCCA-MOT: multimodal Collaboration-Guided Cascade Association Network for 3d multi-object tracking

引用

IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2025年第1期26卷 974-989页

作者： Li, Hui Liu, Hengyuan du, Zhenzhen Chen, Zhong Tao, Ye Qingdao Univ Sci & Technol Dept Data Sci Qingdao 266061 Peoples R China Qingdao Univ Sci & Technol Dept Informat Sci & Technol Qingdao 266061 Peoples R China Southern Illinois Univ Dept Comp Carbondale IL 62901 USA

3d multi-object tracking is an important component of autonomous driving technology. Recent 3d multi-object tracking methods still suffer from issues such as information loss during the fusion of multimodal features, weak discriminative power of the association matrix, and poor robustness of single similarity measure. To address these problems, this paper proposes a multimodal Collaboration-guided Cascade Association network for 3d multi-object tracking (MCCA-MOT). We design a point cloud feature adaptive diffusion fusion module. This module utilizes inverse distance weighting aggregation diffusion technology to address the issue of information loss during the feature fusion process. This enhances the tracking performance of small objects. Secondly, we propose a dynamic sampling feature cooperative fusion module. This module performs fine-grained local-global feature cooperative fusion based on dynamic sampling, enhancing the distinctiveness of object features. It improves the tracking capability of occluded objects. Finally, in the multi-similarity measure-driven cascading association module, we construct a more discriminative association matrix using multiple types of information and design a cascading strategy. This strategy applies different similarity measures at different association stages for objects with ambiguous features. This reduces identity switches and trajectory fragmentation. Extensive experiments on the KITTI dataset demonstrate the superiority of our method in various performance metrics. Our detailed implementations can be obtained at https://***/yuanfuture/MCCA-MOT.

关键词： Three-dimensional displays Accuracy Point cloud compression object tracking Feature extraction Robustness Trajectory object recognition Loss measurement Laser radar 3d multi-object tracking point cloud multimodal fusion data association affinity matrix

来源：评论

学校读者我要写书评

暂无评论

mmMCL3dMOT: multi-Modal Momentum Contrastive Learning for 3d multi-object tracking

引用

IEEE SIGNAL PROCESSING LETTERS 2024年 31卷 1895-1899页

作者： Hong, Ru Yang, Jiming Zhou, Weidian da, Feipeng Southeast Univ Sch Automation Minist Educ Nanjing 210096 Peoples R China Southeast Univ Key Lab Measurement & Control Complex Syst Engn Minist Educ Nanjing Peoples R China

3d multi-object tracking methods utilize object motion, image, and point cloud information to compute similarities between different objects, facilitating cross-frame data association. In this letter, we propose a novel approach called mmMCL3dMOT to calculate object appearance similarity by employing multi-modal momentum contrastive self-supervised learning. We introduce three key techniques. First, a self-supervised training paradigm is adopted, incorporating image, point cloud, and existing 3d detection inputs to enable multi-modal feature extraction without manual annotation. Second, our feature learning approach combines intra-modal and cross-modal feature correspondences within image and point cloud modalities, resulting in more discriminative feature extraction with momentum contrast. Finally, by computing similarity using the multi-modal features and incorporating a robust motion metric, we enable joint cascade reasoning for object association, leading to high-performance 3d MOT. Extensive experiments have demonstrated the significant impact of our method. Moreover, our tracker achieves state-of-the-art (SOTA) performance on both the KITTI and nuScenes datasets.

关键词： Three-dimensional displays Point cloud compression Feature extraction Costs Contrastive learning Training Solid modeling 3d multi-object tracking multi-modal momentum contrast self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Integrated detection and tracking Framework for 3d multi-object tracking in Vehicle-Infrastructure Cooperation

引用

INTERNATIONAL JOURNAL OF AdVANCEd COMPUTER SCIENCE ANd APPLICATIONS 2024年第11期15卷 1228-1237页

作者： Hu, Tao Wang, Ping Wang, Xinhong Tongji Univ Coll Elect & Informat Engn Shanghai Peoples R China

Vehicle-infrastructure cooperative perception has emerged as a promising approach to enhance 3d multi-object tracking by leveraging complementary data from vehicle and infrastructure sensors. However, existing methods face significant challenges, including difficulty in handling occlusions, suboptimal identity association, and inefficiencies in trajectory management, limiting their performance in real-world scenarios. In this paper, we propose a novel vehicle-infrastructure cooperative 3d multi-object tracking framework that addresses these challenges through three key innovations. First, an integrated detection- tracking framework jointly optimizes object detection and tracking, enhancing temporal consistency and reducing errors caused by separately handling the two tasks. Second, the XIOU identity association metric leverages 3d spatial and geometric relationships, ensuring robust object matching even under occlusions. Third, a four-stage cascade matching (FSCM) strategy adaptively manages trajectories by leveraging detection and prediction confidences, enabling accurate tracking in complex environments. Evaluated on the V2X-Seq dataset, our method achieves a MOTH of 57.23 and a MOTP of 74.64, significantly reducing identity switches while ensuring low bandwidth consumption and reliable tracking, highlighting its effectiveness and suitability for real- world deployment.

关键词： Vehicle-infrastructure cooperative perception 3d multi-object tracking XIOU metric four-stage cascade matching integrated detection-tracking framework

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：