Pro2Diff: Proposal Propagation for Multi-Object Tracking via the Diffusion Model

Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate competitive learning between the detection and tracking subtasks, for which, two-stage Tracking-By-Detection (TBD) optimizes the two subtasks individ...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on image processing Jg. 33; S. 6508 - 6520
Hauptverfasser: Liu, Hongmin, Zhang, Canbin, Fan, Bin, Xu, Jinglin
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States IEEE 2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
ISSN:1057-7149, 1941-0042, 1941-0042
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate competitive learning between the detection and tracking subtasks, for which, two-stage Tracking-By-Detection (TBD) optimizes the two subtasks individually, and the single-stage Joint Detection and Tracking (JDT) adjusts the complex network architectures finely in an end-to-end pipeline. In this paper, we propose a new MOT method, i.e., Proposal Propagation via Diffusion Models, called Pro2Diff, which integrates a diffusion model into the proposal propagation in multi-object tracking, focusing on the model training process rather than complex network design. Specifically, using a generative approach, Pro2Diff generates a considerable number of noisy proposals for the tracking image sequence in the forward process, and subsequently, Pro2Diff learns the discrepancies between these noisy proposals and the actual bounding boxes of the tracked objects, gradually optimizing these noisy proposals to obtain the initial sequence of real tracked objects. By introducing the denoising diffusion process into multi-object tracking, we have made three further important findings: 1) Generative methods can effectively handle multi-object tracking tasks; 2) Without the need to modify the model structure, we propose self-conditional proposal propagation to enhance model performance effectively during inference; 3) By adjusting the numbers of proposals and iterations appropriately for different tracking sequences, the optimal performance of the model can be achieved. Extensive experimental results on MOT17 and DanceTrack datasets demonstrate that Pro2Diff outperforms current end-to-end multi-object tracking methods. We achieve 61.9 HOTA on DanceTrack and 57.6 HOTA on MOT17, reaching the competitive result of the JDT approach.
AbstractList Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate competitive learning between the detection and tracking subtasks, for which, two-stage Tracking-By-Detection (TBD) optimizes the two subtasks individually, and the single-stage Joint Detection and Tracking (JDT) adjusts the complex network architectures finely in an end-to-end pipeline. In this paper, we propose a new MOT method, i.e., Proposal Propagation via Diffusion Models, called Pro2Diff, which integrates a diffusion model into the proposal propagation in multi-object tracking, focusing on the model training process rather than complex network design. Specifically, using a generative approach, Pro2Diff generates a considerable number of noisy proposals for the tracking image sequence in the forward process, and subsequently, Pro2Diff learns the discrepancies between these noisy proposals and the actual bounding boxes of the tracked objects, gradually optimizing these noisy proposals to obtain the initial sequence of real tracked objects. By introducing the denoising diffusion process into multi-object tracking, we have made three further important findings: 1) Generative methods can effectively handle multi-object tracking tasks; 2) Without the need to modify the model structure, we propose self-conditional proposal propagation to enhance model performance effectively during inference; 3) By adjusting the numbers of proposals and iterations appropriately for different tracking sequences, the optimal performance of the model can be achieved. Extensive experimental results on MOT17 and DanceTrack datasets demonstrate that Pro2Diff outperforms current end-to-end multi-object tracking methods. We achieve 61.9 HOTA on DanceTrack and 57.6 HOTA on MOT17, reaching the competitive result of the JDT approach.
Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate competitive learning between the detection and tracking subtasks, for which, two-stage Tracking-By-Detection (TBD) optimizes the two subtasks individually, and the single-stage Joint Detection and Tracking (JDT) adjusts the complex network architectures finely in an end-to-end pipeline. In this paper, we propose a new MOT method, i.e., Proposal Propagation via Diffusion Models, called Pro2Diff, which integrates a diffusion model into the proposal propagation in multi-object tracking, focusing on the model training process rather than complex network design. Specifically, using a generative approach, Pro2Diff generates a considerable number of noisy proposals for the tracking image sequence in the forward process, and subsequently, Pro2Diff learns the discrepancies between these noisy proposals and the actual bounding boxes of the tracked objects, gradually optimizing these noisy proposals to obtain the initial sequence of real tracked objects. By introducing the denoising diffusion process into multi-object tracking, we have made three further important findings: 1) Generative methods can effectively handle multi-object tracking tasks; 2) Without the need to modify the model structure, we propose self-conditional proposal propagation to enhance model performance effectively during inference; 3) By adjusting the numbers of proposals and iterations appropriately for different tracking sequences, the optimal performance of the model can be achieved. Extensive experimental results on MOT17 and DanceTrack datasets demonstrate that Pro2Diff outperforms current end-to-end multi-object tracking methods. We achieve 61.9 HOTA on DanceTrack and 57.6 HOTA on MOT17, reaching the competitive result of the JDT approach.Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate competitive learning between the detection and tracking subtasks, for which, two-stage Tracking-By-Detection (TBD) optimizes the two subtasks individually, and the single-stage Joint Detection and Tracking (JDT) adjusts the complex network architectures finely in an end-to-end pipeline. In this paper, we propose a new MOT method, i.e., Proposal Propagation via Diffusion Models, called Pro2Diff, which integrates a diffusion model into the proposal propagation in multi-object tracking, focusing on the model training process rather than complex network design. Specifically, using a generative approach, Pro2Diff generates a considerable number of noisy proposals for the tracking image sequence in the forward process, and subsequently, Pro2Diff learns the discrepancies between these noisy proposals and the actual bounding boxes of the tracked objects, gradually optimizing these noisy proposals to obtain the initial sequence of real tracked objects. By introducing the denoising diffusion process into multi-object tracking, we have made three further important findings: 1) Generative methods can effectively handle multi-object tracking tasks; 2) Without the need to modify the model structure, we propose self-conditional proposal propagation to enhance model performance effectively during inference; 3) By adjusting the numbers of proposals and iterations appropriately for different tracking sequences, the optimal performance of the model can be achieved. Extensive experimental results on MOT17 and DanceTrack datasets demonstrate that Pro2Diff outperforms current end-to-end multi-object tracking methods. We achieve 61.9 HOTA on DanceTrack and 57.6 HOTA on MOT17, reaching the competitive result of the JDT approach.
Author Liu, Hongmin
Zhang, Canbin
Xu, Jinglin
Fan, Bin
Author_xml – sequence: 1
  givenname: Hongmin
  orcidid: 0000-0001-9834-4087
  surname: Liu
  fullname: Liu, Hongmin
  email: hmliu_82@163.com
  organization: School of Intelligence Science and Technology and the Institute of Artificial Intelligence, University of Science and Technology Beijing, Beijing, China
– sequence: 2
  givenname: Canbin
  surname: Zhang
  fullname: Zhang, Canbin
  email: M202210574@xs.ustb.edu.cn
  organization: School of Intelligence Science and Technology and the Institute of Artificial Intelligence, University of Science and Technology Beijing, Beijing, China
– sequence: 3
  givenname: Bin
  orcidid: 0000-0002-1155-467X
  surname: Fan
  fullname: Fan, Bin
  email: bin.fan@ieee.org
  organization: School of Intelligence Science and Technology and the Institute of Artificial Intelligence, University of Science and Technology Beijing, Beijing, China
– sequence: 4
  givenname: Jinglin
  orcidid: 0000-0002-1553-5441
  surname: Xu
  fullname: Xu, Jinglin
  email: xujinglinlove@gmail.com
  organization: School of Intelligence Science and Technology and the Institute of Artificial Intelligence, University of Science and Technology Beijing, Beijing, China
BackLink https://www.ncbi.nlm.nih.gov/pubmed/40030231$$D View this record in MEDLINE/PubMed
BookMark eNp9kTtPwzAUhS0Eog_YGRCKxMKSch07D7Oh8qrUqh3KHNmOU1zSuNgJEv8e9wFCHZh8ZH3n3qtzeui4NrVC6ALDAGNgt_PRbBBBRAeEMpoAHKEuZhSHADQ69hriNEwxZR3Uc24JgGmMk1PUoQAEIoK7aDazJnrQZXkXeLU2jldbwRe80aYOSmODSVs1OpyKpZJNMLdcvut6EXxqHjRvKtiYW7dhJ6ZQ1Rk6KXnl1Pn-7aPXp8f58CUcT59Hw_txKAnNmpCmghQ8YwmPU5JlEos4yTguYiZEqUREith_UiwYVpyWkQRaeJ0pXLJMZoL00c1u7tqaj1a5Jl9pJ1VV8VqZ1uUEp4QwH0bi0esDdGlaW_vrPEWAkYh6uI-u9lQrVqrI11avuP3Kf7LyAOwAaY1zVpW_CIZ8U0fu68g3deT7OrwlObBI3WyDbSzX1X_Gy51RK6X-7EljQikj3_I0lPw
CODEN IIPRE4
CitedBy_id crossref_primary_10_3390_sym17040608
crossref_primary_10_1007_s10462_025_11212_y
Cites_doi 10.1109/TCSVT.2023.3241988
10.1109/CVPR52688.2022.00296
10.1109/CVPR52729.2023.02160
10.1109/tnnls.2023.3274926
10.1007/978-3-030-58621-8_7
10.1109/CVPR42600.2020.01468
10.1145/3581783.3612405
10.1109/TIP.2022.3154286
10.1109/ICCV51070.2023.00951
10.1109/TIP.2022.3165376
10.1109/CVPR52733.2024.01386
10.1109/CVPR46437.2021.01217
10.1109/CVPR52729.2023.01720
10.1007/978-3-030-58548-8_9
10.1109/DATE.2010.5456912
10.1109/TIV.2019.2938110
10.1109/TCSVT.2014.2355692
10.1109/ICCV48922.2021.01297
10.1109/CVPR.2017.394
10.1007/978-3-030-58548-8_28
10.1109/TPAMI.2024.3366234
10.1109/ICASSP49357.2023.10095167
10.1109/ICCV.2019.00103
10.1109/CVPR52688.2022.00857
10.1109/TPAMI.2024.3350049
10.1109/CVPR52729.2023.00289
10.1038/s41598-023-27621-4
10.1109/CVPRW56347.2022.00080
10.1109/CVPRW59228.2023.00296
10.1109/TIP.2020.2993073
10.1109/ICCV51070.2023.00210
10.1109/WACVW54805.2022.00019
10.1109/CVPR52729.2023.02112
10.1049/cje.2020.00.007
10.1109/TPAMI.2024.3382753
10.1007/978-3-319-10602-1_48
10.1109/TIP.2020.3044219
10.1109/CVPR52733.2024.01830
10.1109/CVPR42600.2020.01110
10.1007/978-3-031-20047-2_1
10.1109/TPAMI.2023.3301975
10.1109/ICCV.2019.00409
10.1109/ICCV51070.2023.00930
10.1109/CVPR46437.2021.00387
10.1145/1177352.1177355
10.1007/978-3-030-58452-8_13
10.1109/ICIP.2016.7533003
10.1109/CVPR.2016.90
10.1109/ICCV51070.2023.00910
10.1109/TIP.2022.3163544
10.1109/ICRA48506.2021.9561110
10.1109/TIP.2022.3166278
10.1109/ICRA48506.2021.9561235
10.1109/CVPR42600.2020.00688
10.1109/ICIP.2017.8296962
10.1609/icwsm.v17i1.22160
10.1145/3581783.3612406
10.1109/CVPR52729.2023.00539
10.1109/ISBI53787.2023.10230346
10.1049/cje.2021.00.455
10.1108/RIA-05-2023-0060
10.1109/CVPR52733.2024.02057
10.1109/CVPR52733.2024.00231
10.1007/978-3-031-19812-0_38
10.1109/CVPR52688.2022.02032
10.1109/CVPR52733.2024.01492
ContentType Journal Article
Copyright Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024
Copyright_xml – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2024
DBID 97E
RIA
RIE
AAYXX
CITATION
NPM
7SC
7SP
8FD
JQ2
L7M
L~C
L~D
7X8
DOI 10.1109/TIP.2024.3494600
DatabaseName IEEE Xplore (IEEE)
IEEE All-Society Periodicals Package (ASPP) 1998–Present
IEEE Electronic Library (IEL)
CrossRef
PubMed
Computer and Information Systems Abstracts
Electronics & Communications Abstracts
Technology Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
MEDLINE - Academic
DatabaseTitle CrossRef
PubMed
Technology Research Database
Computer and Information Systems Abstracts – Academic
Electronics & Communications Abstracts
ProQuest Computer Science Collection
Computer and Information Systems Abstracts
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts Professional
MEDLINE - Academic
DatabaseTitleList
PubMed
MEDLINE - Academic
Technology Research Database
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
– sequence: 3
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod fulltext_linktorsrc
Discipline Applied Sciences
Engineering
EISSN 1941-0042
EndPage 6520
ExternalDocumentID 40030231
10_1109_TIP_2024_3494600
10753449
Genre orig-research
Journal Article
GrantInformation_xml – fundername: National Natural Science Foundation of China
  grantid: 62222302; 62373043
  funderid: 10.13039/501100001809
GroupedDBID ---
-~X
.DC
0R~
29I
4.4
53G
5GY
5VS
6IK
97E
AAJGR
AARMG
AASAJ
AAWTH
ABAZT
ABFSI
ABQJQ
ABVLG
ACGFO
ACGFS
ACIWK
AENEX
AETIX
AGQYO
AGSQL
AHBIQ
AI.
AIBXA
AKJIK
AKQYR
ALLEH
ALMA_UNASSIGNED_HOLDINGS
ASUFR
ATWAV
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CS3
DU5
E.L
EBS
EJD
F5P
HZ~
H~9
ICLAB
IFIPE
IFJZH
IPLJI
JAVBF
LAI
M43
MS~
O9-
OCL
P2P
RIA
RIE
RNS
TAE
TN5
VH1
AAYXX
CITATION
NPM
RIG
7SC
7SP
8FD
JQ2
L7M
L~C
L~D
7X8
ID FETCH-LOGICAL-c348t-47b3da896a57388c1b568a1d59bbfeb23d58c141b91ea4f2c04db918e1f98c8b3
IEDL.DBID RIE
ISICitedReferencesCount 2
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001360426000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1057-7149
1941-0042
IngestDate Thu Oct 02 11:17:51 EDT 2025
Mon Jun 30 10:16:19 EDT 2025
Mon Jul 21 06:06:31 EDT 2025
Sat Nov 29 03:34:45 EST 2025
Tue Nov 18 21:35:28 EST 2025
Wed Aug 27 03:06:43 EDT 2025
IsPeerReviewed true
IsScholarly true
Language English
License https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html
https://doi.org/10.15223/policy-029
https://doi.org/10.15223/policy-037
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c348t-47b3da896a57388c1b568a1d59bbfeb23d58c141b91ea4f2c04db918e1f98c8b3
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ORCID 0000-0002-1155-467X
0000-0002-1553-5441
0000-0001-9834-4087
PMID 40030231
PQID 3130932417
PQPubID 85429
PageCount 13
ParticipantIDs proquest_journals_3130932417
ieee_primary_10753449
crossref_primary_10_1109_TIP_2024_3494600
pubmed_primary_40030231
crossref_citationtrail_10_1109_TIP_2024_3494600
proquest_miscellaneous_3173399416
PublicationCentury 2000
PublicationDate 20240000
2024-00-00
20240101
PublicationDateYYYYMMDD 2024-01-01
PublicationDate_xml – year: 2024
  text: 20240000
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
– name: New York
PublicationTitle IEEE transactions on image processing
PublicationTitleAbbrev TIP
PublicationTitleAlternate IEEE Trans Image Process
PublicationYear 2024
Publisher IEEE
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Publisher_xml – name: IEEE
– name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
References ref13
ref57
ref12
Leal-Taixé (ref45) 2015
ref56
ref15
ref59
ref14
ref53
ref52
ref11
ref55
ref10
ref54
ref17
ref16
ref19
ref18
Baranchuk (ref61) 2021
ref51
ref50
ref46
Chen (ref42) 2022
ref48
ref41
ref85
ref44
ref88
ref43
ref87
Ge (ref35) 2021
Ho (ref36); 33
ref49
ref8
ref7
ref9
ref4
ref3
ref5
Ho (ref39) 2022
ref82
ref81
ref40
ref84
ref83
Wu (ref2) 2023
ref80
ref79
ref78
ref31
Zhang (ref47) 2020
ref30
ref74
ref33
ref77
ref32
Sun (ref86) 2020
ref1
Wang (ref34) 2020
Gu (ref69) 2022
Cao (ref89) 2022
ref70
ref73
ref72
Song (ref37) 2020
ref24
ref68
ref23
ref67
ref26
ref20
Chen (ref60) 2022
ref64
ref63
ref22
ref66
ref21
ref65
Yang (ref71) 2023
Lu (ref38); 35
ref27
Meinhardt (ref28) 2021
ref29
Wu (ref62) 2022
Zhu (ref76) 2020
Su (ref6) 2025; 34
Fang (ref58)
Nichol (ref75); 139
Milan (ref25) 2016
References_xml – ident: ref1
  doi: 10.1109/TCSVT.2023.3241988
– ident: ref13
  doi: 10.1109/CVPR52688.2022.00296
– year: 2022
  ident: ref42
  article-title: DiffusionDet: Diffusion model for object detection
  publication-title: arXiv:2211.09788
– year: 2020
  ident: ref86
  article-title: TransTrack: Multiple object tracking with transformer
  publication-title: arXiv:2012.15460
– ident: ref59
  doi: 10.1109/CVPR52729.2023.02160
– ident: ref20
  doi: 10.1109/tnnls.2023.3274926
– ident: ref31
  doi: 10.1007/978-3-030-58621-8_7
– ident: ref32
  doi: 10.1109/CVPR42600.2020.01468
– ident: ref56
  doi: 10.1145/3581783.3612405
– ident: ref80
  doi: 10.1109/TIP.2022.3154286
– ident: ref43
  doi: 10.1109/ICCV51070.2023.00951
– year: 2023
  ident: ref71
  article-title: Boosting human-object interaction detection with Text-to-Image diffusion model
  publication-title: arXiv:2305.12252
– ident: ref48
  doi: 10.1109/TIP.2022.3165376
– year: 2021
  ident: ref61
  article-title: Label-efficient semantic segmentation with diffusion models
  publication-title: arXiv:2112.03126
– year: 2015
  ident: ref45
  article-title: MOTChallenge 2015: Towards a benchmark for multi-target tracking
  publication-title: arXiv:1504.01942
– ident: ref16
  doi: 10.1109/CVPR52733.2024.01386
– ident: ref49
  doi: 10.1109/CVPR46437.2021.01217
– ident: ref54
  doi: 10.1109/CVPR52729.2023.01720
– ident: ref30
  doi: 10.1007/978-3-030-58548-8_9
– ident: ref53
  doi: 10.1109/DATE.2010.5456912
– volume: 34
  start-page: 1
  issue: 1
  year: 2025
  ident: ref6
  article-title: An improved YOLOv7-tiny algorithm for vehicle and pedestrian detection with occlusion in autonomous driving
  publication-title: CJE
– start-page: 16716
  volume-title: Proc. NeurIPS
  ident: ref58
  article-title: Structural pruning for diffusion models
– ident: ref4
  doi: 10.1109/TIV.2019.2938110
– ident: ref11
  doi: 10.1109/TCSVT.2014.2355692
– ident: ref84
  doi: 10.1109/ICCV48922.2021.01297
– ident: ref23
  doi: 10.1109/CVPR.2017.394
– ident: ref29
  doi: 10.1007/978-3-030-58548-8_28
– year: 2022
  ident: ref39
  article-title: Imagen video: High definition video generation with diffusion models
  publication-title: arXiv:2210.02303
– volume: 33
  start-page: 6840
  volume-title: Proc. NIPS
  ident: ref36
  article-title: Denoising diffusion probabilistic models
– ident: ref12
  doi: 10.1109/TPAMI.2024.3366234
– ident: ref66
  doi: 10.1109/ICASSP49357.2023.10095167
– ident: ref9
  doi: 10.1109/ICCV.2019.00103
– ident: ref88
  doi: 10.1109/CVPR52688.2022.00857
– ident: ref21
  doi: 10.1109/TPAMI.2024.3350049
– volume: 35
  start-page: 5775
  volume-title: Proc. NeurIPS
  ident: ref38
  article-title: DPM-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps
– ident: ref72
  doi: 10.1109/CVPR52729.2023.00289
– year: 2021
  ident: ref35
  article-title: YOLOX: Exceeding YOLO series in 2021
  publication-title: arXiv:2107.08430
– ident: ref63
  doi: 10.1038/s41598-023-27621-4
– ident: ref65
  doi: 10.1109/CVPRW56347.2022.00080
– ident: ref67
  doi: 10.1109/CVPRW59228.2023.00296
– ident: ref79
  doi: 10.1109/TIP.2020.2993073
– ident: ref73
  doi: 10.1109/ICCV51070.2023.00210
– year: 2020
  ident: ref37
  article-title: Denoising diffusion implicit models
  publication-title: arXiv:2010.02502
– year: 2020
  ident: ref76
  article-title: Deformable DETR: Deformable transformers for end-to-end object detection
  publication-title: arXiv:2010.04159
– ident: ref85
  doi: 10.1109/WACVW54805.2022.00019
– ident: ref52
  doi: 10.1109/CVPR52729.2023.02112
– year: 2016
  ident: ref25
  article-title: MOT16: A benchmark for multi-object tracking
  publication-title: arXiv:1603.00831
– year: 2022
  ident: ref69
  article-title: DiffusionInst: Diffusion model for instance segmentation
  publication-title: arXiv:2212.02773
– ident: ref7
  doi: 10.1049/cje.2020.00.007
– ident: ref40
  doi: 10.1109/TPAMI.2024.3382753
– ident: ref78
  doi: 10.1007/978-3-319-10602-1_48
– year: 2022
  ident: ref89
  article-title: Observation-centric SORT: Rethinking SORT for robust multi-object tracking
  publication-title: arXiv:2203.14360
– year: 2020
  ident: ref47
  article-title: FairMOT: On the fairness of detection and re-identification in multiple object tracking
  publication-title: arXiv:2004.01888
– ident: ref87
  doi: 10.1109/TIP.2020.3044219
– ident: ref41
  doi: 10.1109/CVPR52733.2024.01830
– ident: ref33
  doi: 10.1109/CVPR42600.2020.01110
– ident: ref46
  doi: 10.1007/978-3-031-20047-2_1
– ident: ref81
  doi: 10.1109/TPAMI.2023.3301975
– ident: ref24
  doi: 10.1109/ICCV.2019.00409
– volume: 139
  start-page: 8162
  volume-title: Proc. 38th Int. Conf. Mach. Learn.
  ident: ref75
  article-title: Improved denoising diffusion probabilistic models
– ident: ref70
  doi: 10.1109/ICCV51070.2023.00930
– ident: ref83
  doi: 10.1109/CVPR46437.2021.00387
– ident: ref5
  doi: 10.1145/1177352.1177355
– year: 2020
  ident: ref34
  article-title: Joint object detection and multi-object tracking with graph neural networks
  publication-title: arXiv:2006.13164
– ident: ref51
  doi: 10.1007/978-3-030-58452-8_13
– ident: ref27
  doi: 10.1109/ICIP.2016.7533003
– ident: ref77
  doi: 10.1109/CVPR.2016.90
– ident: ref14
  doi: 10.1109/ICCV51070.2023.00910
– ident: ref19
  doi: 10.1109/TIP.2022.3163544
– ident: ref82
  doi: 10.1109/ICRA48506.2021.9561110
– ident: ref18
  doi: 10.1109/TIP.2022.3166278
– ident: ref8
  doi: 10.1109/ICRA48506.2021.9561235
– ident: ref26
  doi: 10.1109/CVPR42600.2020.00688
– ident: ref22
  doi: 10.1109/ICIP.2017.8296962
– ident: ref68
  doi: 10.1609/icwsm.v17i1.22160
– ident: ref55
  doi: 10.1145/3581783.3612406
– ident: ref74
  doi: 10.1109/CVPR52729.2023.00539
– ident: ref64
  doi: 10.1109/ISBI53787.2023.10230346
– ident: ref3
  doi: 10.1049/cje.2021.00.455
– ident: ref10
  doi: 10.1108/RIA-05-2023-0060
– ident: ref17
  doi: 10.1109/CVPR52733.2024.02057
– ident: ref15
  doi: 10.1109/CVPR52733.2024.00231
– ident: ref50
  doi: 10.1007/978-3-031-19812-0_38
– ident: ref44
  doi: 10.1109/CVPR52688.2022.02032
– year: 2023
  ident: ref2
  article-title: Towards video anomaly retrieval from video anomaly detection: New benchmarks and model
  publication-title: arXiv:2307.12545
– year: 2022
  ident: ref62
  article-title: MedSegDiff: Medical image segmentation with diffusion probabilistic model
  publication-title: arXiv:2211.00611
– year: 2021
  ident: ref28
  article-title: TrackFormer: Multi-object tracking with transformers
  publication-title: arXiv:2101.02702
– ident: ref57
  doi: 10.1109/CVPR52733.2024.01492
– year: 2022
  ident: ref60
  article-title: A generalist framework for panoptic segmentation of images and videos
  publication-title: arXiv:2210.06366
SSID ssj0014516
Score 2.4503894
Snippet Multi-object tracking (MOT) aims to estimate the bounding boxes and ID labels of objects in videos. The challenging issue in this task is to alleviate...
SourceID proquest
pubmed
crossref
ieee
SourceType Aggregation Database
Index Database
Enrichment Source
Publisher
StartPage 6508
SubjectTerms Boxes
diffusion models
Multi-object tracking
Multiple target tracking
Network design
Object recognition
Optimization
Propagation
proposal propagation
Proposals
Title Pro2Diff: Proposal Propagation for Multi-Object Tracking via the Diffusion Model
URI https://ieeexplore.ieee.org/document/10753449
https://www.ncbi.nlm.nih.gov/pubmed/40030231
https://www.proquest.com/docview/3130932417
https://www.proquest.com/docview/3173399416
Volume 33
WOSCitedRecordID wos001360426000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVIEE
  databaseName: IEEE Electronic Library (IEL)
  customDbUrl:
  eissn: 1941-0042
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0014516
  issn: 1057-7149
  databaseCode: RIE
  dateStart: 19920101
  isFulltext: true
  titleUrlDefault: https://ieeexplore.ieee.org/
  providerName: IEEE
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Lb9swDCbaooftsLZZ17prAw3YZQe1cSTr0dvQB9pL5kMG5GZIsjwEKJIij_7-kbIT5NIBuwm2ZAsiaX4kTRLguwtKDZxQPA_Rchmc5R6BNc-Fc4UZhtw7mZpN6NHITCa27JLVUy5MjDH9fBavaZhi-fU8rMlVhhKO4FpKuw_7Wqs2WWsbMqCOsym0WWiuEfdvYpIDezN-LtESHMprqsWiKJltRwelpirv48ukZx6P_nOHx_CpA5TsZ8sBJ7AXZz046sAl60R32YOPO5UHP0NZLubD-2nT3LKS-iQs8RE0cH8SpRhCWZZyc_kvT54ahjotkFedvU0dQ9DIaPGaXG2M2qm9nMLvx4fx3RPvmivwIKRZcam9qJ2xyhVaGINEKZRxeV1Y7xs0t0Vd4EWZe5tHJ5thGMgaxybmjTXBePEFDmbzWTwH5lETFg5xmHAWFyjrGzSihJBRaDwPlcHN5rir0FUepwYYL1WyQAa2QgJVRKCqI1AGP7YrXtuqG_-Ye0p02JnXkiCDyw1Jq04ul5XIKfKLqEVn8G17GyWKwiRuFudrmqMFwjZEqhmctaywfbikjyJC4ot3XvoVPtDeWh_NJRysFut4BYfhbTVdLvrIthPTT2z7F5Af5iM
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LT9wwEB4VWgl6KJQCTXnUSFx68JLETmL3VtEiEHTJYZG4RbbjoJXQLtoHv58ZJ7vaC0jcrMROLH-e-JuZzAzAqXF5HhuR88R5zaUzmlsk1jwRxmQqdYk1MhSbKPp9dX-vyy5YPcTCeO_Dz2e-R83gy6_Hbk6mMpRwJNdS6jX4mEmZxm241tJpQDVng3MzK3iBzH_hlYz12eCqRF0wlT3KxpJTONvKKRTKqrzOMMNJc7H1zjluw5eOUrI_7R74Ch_8aAe2OnrJOuGd7sDnldyD36AsJ-P077BpfrOSKiVM8RHUMA8BK4ZkloXoXH5ryVbD8FRzZFdnz0PDkDYyGjwnYxujgmqPu3B38W9wfsm78grcCalmXBZW1Ebp3GSFUAphyXJlkjrT1jaocIs6w4sysTrxRjapi2WNbeWTRiunrNiD9dF45L8Ds3gWZgaZmDAaB-TaNqhGCSG9KHA98gjOFstduS73OJXAeKyCDhLrCgGqCKCqAyiCX8sRT23ejTf67hIOK_1aCCI4XEBadZI5rURCvl_kLUUEJ8vbKFPkKDEjP55Tn0IgcUOuGsF-uxWWD5f0WURS_OOVl_6EjcvB_5vq5qp_fQCbNM_WYnMI67PJ3B_BJ_c8G04nx2HzvgBDkOiC
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Pro2Diff%3A+Proposal+Propagation+for+Multi-Object+Tracking+via+the+Diffusion+Model&rft.jtitle=IEEE+transactions+on+image+processing&rft.au=Liu%2C+Hongmin&rft.au=Zhang%2C+Canbin&rft.au=Fan%2C+Bin&rft.au=Xu%2C+Jinglin&rft.date=2024&rft.issn=1057-7149&rft.eissn=1941-0042&rft.volume=33&rft.spage=6508&rft.epage=6520&rft_id=info:doi/10.1109%2FTIP.2024.3494600&rft.externalDBID=n%2Fa&rft.externalDocID=10_1109_TIP_2024_3494600
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1057-7149&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1057-7149&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1057-7149&client=summon