Filters








28 Hits in 1.1 sec

TransTrack: Multiple Object Tracking with Transformer [article]

Peize Sun, Jinkun Cao, Yi Jiang, Rufeng Zhang, Enze Xie, Zehuan Yuan, Changhu Wang, Ping Luo
2021 arXiv   pre-print
In this work, we propose TransTrack, a simple but efficient scheme to solve the multiple object tracking problems.  ...  We expect TransTrack to provide a novel perspective for multiple object tracking. The code is available at: .  ...  According to the number of objects to track, the task of object tracking is divided into Single Object Tracking (SOT) and Multiple Object Tracking (MOT).  ... 
arXiv:2012.15460v2 fatcat:pluunsjuyfbnjam5wcok2twnoq

MOTR: End-to-End Multiple-Object Tracking with Transformer [article]

Fangao Zeng, Bin Dong, Yuang Zhang, Tiancai Wang, Xiangyu Zhang, Yichen Wei
2022 arXiv   pre-print
Temporal modeling of objects is a key challenge in multiple-object tracking (MOT). Existing methods track by associating detections through motion-based and appearance-based similarity heuristics.  ...  Track query is transferred and updated frame-by-frame to perform iterative prediction over time. We propose tracklet-aware label assignment to train track queries and newborn object queries.  ...  Limitations MOTR, an online tracker, achieves end-to-end multiple-object tracking.  ... 
arXiv:2105.03247v3 fatcat:holt3fbwlfcphg6zk5i5xx5owy

TransCenter: Transformers with Dense Representations for Multiple-Object Tracking [article]

Yihong Xu, Yutong Ban, Guillaume Delorme, Chuang Gan, Daniela Rus, Xavier Alameda-Pineda
2022 arXiv   pre-print
Despite this wave, building an accurate and efficient multiple-object tracking (MOT) method with transformers is not a trivial task.  ...  Inspired by recent research, we propose TransCenter, a transformer-based MOT architecture with dense representations for accurately tracking all the objects while keeping a reasonable runtime.  ...  CONCLUSION In this paper, we introduce TransCenter, a novel transformer-based architecture for multiple-object tracking.  ... 
arXiv:2103.15145v3 fatcat:4fj2f4vzpnbmhaw6lzf7g6zpfq

Joint Spatial-Temporal and Appearance Modeling with Transformer for Multiple Object Tracking [article]

Peng Dai and Yiqiang Feng and Renliang Weng and Changshui Zhang
2022 arXiv   pre-print
The recent trend in multiple object tracking (MOT) is heading towards leveraging deep learning to boost the tracking performance.  ...  In this paper, we propose a novel solution named TransSTAM, which leverages Transformer to effectively model both the appearance features of each object and the spatial-temporal relationships among objects  ...  Introduction Multiple object tracking (MOT) in videos is an important problem in many application domains.  ... 
arXiv:2205.15495v1 fatcat:eqigsys2ffbq5ozag36eowiojy

AnimalTrack: A Large-scale Benchmark for Multi-Animal Tracking in the Wild [article]

Libo Zhang, Junyuan Gao, Zhen Xiao, Heng Fan
2022 arXiv   pre-print
On average, each sequence comprises of 33 target objects for tracking. In order to ensure high quality, every frame in AnimalTrack is manually labeled with careful inspection and refinement.  ...  We hope that AnimalTrack together with evaluation and analysis will foster further progress on multi-animal tracking.  ...  TransTrack utilizes the query-key mechanism in Transformer for multi-object tracking. The competitive performance of TransTrack shows the potential of Transformer for MOT.  ... 
arXiv:2205.00158v1 fatcat:a3qux6crazbvvlbo2urzgoz4ge

TransMOT: Spatial-Temporal Graph Transformer for Multiple Object Tracking [article]

Peng Chu, Jiang Wang, Quanzeng You, Haibin Ling, Zicheng Liu
2021 arXiv   pre-print
Tracking multiple objects in videos relies on modeling the spatial-temporal interactions of the objects.  ...  TransMOT effectively models the interactions of a large number of objects by arranging the trajectories of the tracked objects as a set of sparse weighted graphs, and constructing a spatial graph transformer  ...  Conclusion We proposed a novel Spatial-Temporal Graph Transformer for multi-object tracking (TransMOT) with Transformers.  ... 
arXiv:2104.00194v2 fatcat:jhimxr4i4jffrnbqs7itm72r2u

TdmTracker: Multi-Object Tracker Guided by Trajectory Distribution Map

Yuxuan Gao, Xiaohui Gu, Qiang Gao, Runmin Hou, Yuanlong Hou
2022 Electronics  
With the great progress of object detection, some detection-based multiple object tracking (MOT) paradigms begin to emerge, including tracking-by-detection, joint detection and tracking, and attention  ...  Some of the transformer-based end-to-end methods introducing track queries to detect targets have achieved good results. Self-attention and track query of these methods has given us some inspiration.  ...  TransTrack and TrackFormer introduce the transformer architecture.  ... 
doi:10.3390/electronics11071010 fatcat:qfqzfdovifaohdgsjrqlycwaxu

MeMOT: Multi-Object Tracking with Memory [article]

Jiarui Cai, Mingze Xu, Wei Li, Yuanjun Xiong, Wei Xia, Zhuowen Tu, Stefano Soatto
2022 arXiv   pre-print
the core information from the memory for each tracked object; and 3) Memory Decoding that solves the object detection and data association tasks simultaneously for multi-object tracking.  ...  We propose an online tracking algorithm that performs the object detection and data association under a common framework, capable of linking objects after a long time span.  ...  For instance, occlusion in visual multiple object tracking is clearly non-linear and non-Gaussian.  ... 
arXiv:2203.16761v1 fatcat:6arhpggwivgybdnhehyj2jm7se

Global Tracking Transformers [article]

Xingyi Zhou, Tianwei Yin, Vladlen Koltun, Philipp Krähenbühl
2022 arXiv   pre-print
Our global tracking transformer does not require intermediate pairwise grouping or combinatorial association, and can be jointly trained with an object detector.  ...  The core component is a global tracking transformer that operates on objects from all frames in the sequence.  ...  TransTrack [40] uses features from historical tracks as queries, but associates objects based on updated bounding box locations.  ... 
arXiv:2203.13250v2 fatcat:h43zxsqbbfcxtjjdt76mb5heu4

Instance Sequence Queries for Video Instance Segmentation with Transformers

Zhujun Xu, Damien Vivet
2021 Sensors  
In this work, we propose a frame-to-frame method built upon transformers.  ...  On TITAN Xp GPU, our method achieves a competitive 34.4% mAP at 33.5 FPS with ResNet-50 and 35.5% mAP at 26.6 FPS with ResNet-101 on the Youtube-VIS dataset.  ...  TransTrack [20] constructs a query-key pipeline to solve the multiple-object tracking (MOT) task.  ... 
doi:10.3390/s21134507 pmid:34209420 fatcat:6jwfqgjkijcoppc2asxswsykh4

Transformers Meet Visual Learning Understanding: A Comprehensive Review [article]

Yuting Yang, Licheng Jiao, Xu Liu, Fang Liu, Shuyuan Yang, Zhixi Feng, Xu Tang
2022 arXiv   pre-print
The former mainly includes image classification, object detection, and image segmentation. The latter contains object tracking and video classification.  ...  Dynamic attention mechanism and global modeling ability make Transformer show strong feature learning ability. In recent years, Transformer has become comparable to CNNs methods in computer vision.  ...  Swin-Transformer Tracker, namely SwinTrack, is proposed in [124] 2) Multi-object tracking: Transformer-based tracking methods, including TransTrack [118] , TrackFormer [117] , TrSiam and TrDiMP [  ... 
arXiv:2203.12944v1 fatcat:h2kgxfnqqvcbfelvpnteqpytcu

TR-MOT: Multi-Object Tracking by Reference [article]

Mingfei Chen, Yue Liao, Si Liu, Fei Wang, Jenq-Neng Hwang
2022 arXiv   pre-print
Multi-object Tracking (MOT) generally can be split into two sub-tasks, i.e., detection and association.  ...  each object among frames.  ...  MOT aims to estimate the trajectory of objects with the same identity based on a video sequence that contains multiple objects.  ... 
arXiv:2203.16621v1 fatcat:32a7qn3rsbb5rcu4becdjeetke

Detection Recovery in Online Multi-Object Tracking with Sparse Graph Tracker [article]

Jeongseok Hyun, Myunggu Kang, Dongyoon Wee, Dit-Yan Yeung
2022 arXiv   pre-print
Joint object detection and online multi-object tracking (JDT) methods have been proposed recently to achieve one-shot tracking.  ...  The missed detections affect not only detection performance but also tracking performance due to inconsistent tracklets.  ...  GSDT [46] , CorrTracker [45] , and TransTrack [43] propose using GNN [32] , correlation layer [15] , and transformer [66] to enhance the features for detection and tracking, respectively.  ... 
arXiv:2205.00968v1 fatcat:4ygmw263zvdmrmkcevlji2mfqa

End-to-End Video Text Spotting with Transformer [article]

Weijia Wu, Debing Zhang, Ying Fu, Chunhua Shen, Hong Zhou, Yuanqiang Cai, Ping Luo
2022 arXiv   pre-print
., detecting text in individual images, recognizing localized text, tracking text streams with post-processing to generate final results.  ...  In this paper, rooted in Transformer sequence modeling, we propose a simple, but effective end-to-end video text DEtection, Tracking, and Recognition framework (TransDETR).  ...  TransTrack [34] track object by applying object features from the previous frame as a query of the current frame and introduces a set of learned object queries to enable detecting new-coming objects.  ... 
arXiv:2203.10539v1 fatcat:xsrdpd7h45ae7lxc2h6g7m2cra

REAL-TIME DEEP NEURAL NETWORKS FOR MULTIPLE OBJECT TRACKING AND SEGMENTATION ON MONOCULAR VIDEO

I. Basharov, D. Yudin
2021 The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences  
The paper is devoted to the task of multiple objects tracking and segmentation on monocular video, which was obtained by the camera of unmanned ground vehicle.  ...  The authors proposed an approach based on combining the modern SOLOv2 instance segmentation model, a neural network model for embedding generation for each found object, and a modified Hungarian tracking  ...  INTRODUCTION Multiple object tracking (MOT) task is very important for a large number of applications.  ... 
doi:10.5194/isprs-archives-xliv-2-w1-2021-15-2021 fatcat:qf7fsx4tynfmpc6kzvm36bgi5q
« Previous Showing results 1 — 15 out of 28 results