VOT2021 Trackers repository

This is an incomplete collection of trackers submitted to VOT2021. The collection contains only those submissions to VOT2021 for which we were able to obtain explicit permission from the authors (this was indicated by the authors during results submission process). In most cases the code is provided as it was submitted, in other cases some authors asked us to point the link to their own repositories. The VOT committee does not take any responsibility for the state of code/binaries of the trackers available here. For any additional information regarding individual trackers, please contact the corresponding authors directly.

Each tracker is referred as NAME (Appendix Nr NN), where NAME is the name given to the tracker and NN is the tracker number referred on the VOT paper.

If you have any question or suggestion, please contact us on the support forum.

Short Term (ST) Trackers

• RPT: Learning Point Set Representation for Siamese Visual Tracking (RPT) - Appendix A.1

• Learning Discriminative Model Prediction for Tracking (DiMP) - Appendix A.2

• Learning What to Learn for Video Object Segmentation (LWL) - Appendix A.3

• Tracking by Student FUSing Teachers (TRASFUSTm) - Appendix A.4

• Learning Spatio-Appearance Memory Network for High-Performance Visual Tracking (SAMN) - Appendix A.5

• SAMN_DiMP (SAMN_DiMP) - Appendix A.6

• Siamse IOU and Occlusion aware Network for Tracking (SION) - Appendix A.7

• Visual Tracking via Adversarial Information Learning - VITAL++ (VITAL++) - Appendix A.8

• Ensemble correlation filter tracking based on a temporal confidence learning (TCLCF) - Appendix A.9

• RPT_AR: Accurate Point Set Representation Tracking by Alpha-Refine(RPT_AR) - Appendix A.10

• SiamBAN-ACM (ACM) - Appendix A.11

• SiamUSC:Uncertainty-aware Semantic Consistency Siamese Tracker (SiamUSC) - Appendix A.12

• Deep Convolutional Descriptor Aggregation Tracker with Accurate Refine Module (DCDAAR) - Appendix A.13

• DeepMix: Online Auto Data Augmentation for Robust Visual Object Tracking (deepmix) - Appendix A.14

• Alpha-Refine (AlphaRef) - Appendix A.15 The complete code and trained models of Alpha-Refine have been released at (github.com/MasterBin-IIAU/AlphaRefine)

• ReptileFPN (ReptileFPN) - Appendix A.16

• STARK for the Real-Time Challenge (transformer) (STARK_RT) - Appendix A.17

• Transformer Tracking (TransT) - Appendix A.18

• fRPT: fast Representative Points based Tracker (fRPT) - Appendix A.19

• SiamUSCPlus: Online Uncertainty-aware Semantic Consistency Siamese Tracker (SiamUSCP) - Appendix A.20

• Siamese Tracker with Template Enhancement and Mask Generation (SiamEM_R) - Appendix A.21

• Multi-Template Transformer Tracking (TransT_M) - Appendix A.22

• NullSpaceRDAR (NSpaceRDAR) - Appendix A.23

• Fast TREG++: fast target transformed regression and segmentation for accurate tracking (F_TregPlus) - Appendix A.24

• TREG++: target transformed regression and segmentation for accurate tracking (TregPlus) - Appendix A.25

• Dual-branch Transformer Tracker with No Convolutional Neural Networks (DualTFR) - Appendix A.26

• Dual-branch Transformer Tracker with No Convolutional Neural Networks (DualTFRst) - Appendix A.27

• CFRPT:Customized Feature Representation for Siamese Visual Tracking (CFRPT) - Appendix A.28

• Visual tracking via Fast Saliency-guided Continuous Correlation Filters (FSC2F) - Appendix A.29

• Discriminative Sing-Shot Segmentation Tracker (D3S) - Appendix A.30

• Discriminative Sing-Shot Segmentation Tracker v2 (D3Sv2) - Appendix A.31

• Dual-branch Transformer Tracker with No Convolutional Neural Networks-online version (DualTFRon) - Appendix A.32

• Refined Transformer Tracker (RTT) - Appendix A.33

• TRATMask: Tracking by Attention Using Spatio-Temporal Features (TRATMask) - Appendix A.34

• KeepTrack (keep_track) - Appendix A.35

• RPTMask (RPTMask) - Appendix A.36

• [Local-Global Tracking tracker (LGT)] - Appendix A.37

• [ANT (ANT)] - Appendix A.38

• [ATOM tracker with Alpha refine post-processing step (AR_ATOM)] - Appendix A.39

• DiMP50 tracker with Alpha refine post-processing step (AR_DiMP-50) - Appendix A.40

• Know your surrondings tracker with Alpha refine post-processing step (AR_KYS) - Appendix A.41

• PrDiMP50 tracker with Alpha refine post-processing step (AR_PrDiMP-50) - Appendix A.42

• SuperDiMP50 tracker with Alpha refine post-processing step (AR_SuperDiMP-50) - Appendix A.43

• Scale adaptive mean shift (ASMS) - Appendix A.44

• Accurate Tracking by Overlap Maximization (ATOM) - Appendix A.45

• [Discriminative Correlation Filter with Channel and Spatial Reliability (CSRDCF)] - Appendix A.46

• Kernelized Correlation Filter (KCF) - Appendix A.47

• [Know your surrondings tracker (KYS)] - Appendix A.48

• [L1APG] - Appendix A.49

• Learning what to learn tracker with Box2Seg head (LWL_B2S) - Appendix A.50

• PrDiMP-50 (PrDiMP-50) - Appendix A.51

• [SiameseFC-AlexNet (SiamFc)] - Appendix A.52

• [VOS SOTA method (STM)] - Appendix A.53

Long Term (LT) Trackers

• Fusing Complementary Trackers for Long-term Visual Tracking (mlpLT) - Appendix B.1

• Siamse IOU and Occlusion aware Network for Tracking (SION_LT) - Appendix B.2

• Siam R-CNN (SiamRCNN) - Appendix B.3

• [SLOT (SLOT)] - Appendix B.4 (This tracker is under submission and will be available upon acceptance)

• DiMP Strengthened STARK for Longterm Tracking (STARK_RGBD_LT) - Appendix B.5

• Target-driven Inference for Deep Video Object Tracking (TDIOT) - Appendix B.6

• STARK for the Long-Term Challenge (STARK_LT) - Appendix B.7

• Switch-Refine Tracking Framework (RincTrack) - Appendix B.8

• KeepTrack (keep_track_lt) - Appendix B.9

• A More Concise Long-Term Tracker with Meta-updater (SuperD_MU) - Appendix B.10

• [SuperDiMP] - Appendix B.11

• LT_DSE - Appendix B.12

• A Baseline LongTerm Tracker with MetaUpdater (LTMU B) - Appendix B.13

RGB-D Trackers

• Spatio-Temporal Transformer with Correlation for RGBD Visual Tracking (sttc_rgbd) - Appendix C.1

• STARK for the RGBD challenge (STARK_RGBD) - Appendix C.2

• Spot-Light Masking feature enhanced DiMP for RGBD Tracking (SLMD) - Appendix C.3

• Towards Accurate RGB-D Tracking by Local and Global Detection (TALGD) - Appendix C.4

• Object tracking based on deep information fusion and consistency constraint update (DRefine) - Appendix C.5