Skip to content

Latest commit

 

History

History
87 lines (76 loc) · 7.97 KB

SOT_PaperList.md

File metadata and controls

87 lines (76 loc) · 7.97 KB

2021年

CVPR

  • SiamGAT:Graph Attention Tracking[paper][code]
  • Alpha-Refine:Boosting Tracking Performance by Precise Bounding Box Estimation[paper][code]
  • LightTrack:Finding Lightweight Neural Networks for Object Tracking via One-Shot Architecture Search[paper][code]
  • STMTrack:Template-free Visual Tracking with Space-time Memory Networks[paper][code]
  • TNL2K:Towards More Flexible and Accurate Object Tracking with Natural Language: Algorithms and Benchmark[paper]
  • TransT:Transformer Tracking[paper][code]
  • TransformerTrack:Transformer Meets Tracker: Exploiting Temporal Context for Robust Visual Tracking[paper][code]
  • SNLT:Siamese Natural Language Tracker: Tracking by Natural Language Descriptions with Siamese Trackers[paper][code]
  • IoU Attack:Towards Temporally Coherent Black-Box Adversarial Attack for Visual Object Tracking[paper][code]
  • RE-SiamNets:Rotation Equivariant Siamese Networks for Tracking[paper][code]
  • Progressive Unsupervised Learning for Visual Object Tracking[paper]
  • Learning To Filter: Siamese Relation Network for Robust Tracking[paper][code]
  • CapsuleRRT: Relationships-aware Regression Tracking via Capsules[paper]

2020年

CVPR

  • SiamCAR: Siamese Fully Convolutional Classification and Regression for Visual Tracking [paper][code]
  • SiamBAN: Siamese Box Adaptive Network for Visual Tracking. [paper][code]
  • D3S – A Discriminative Single Shot Segmentation Tracker. [paper][code]
  • MAML: Tracking by Instance Detection: A Meta-Learning Approach. [paper]
  • SiamAttn: Deformable Siamese Attention Networks for Visual Object Tracking.[paper]
  • CGACD: Correlation-Guided Attention for Corner Detection Based Visual Tracking.
  • Siam R-CNN: Visual Tracking by Re-Detection.[code]

''' 将Faster-RCNN 结合到 Siamese系列跟踪,用RPN网络提取出ROI，将当前帧所有ROI与第一帧的GT的ROI cat进行re-detection，选定得分较高的boxes，再与上一帧的boxes两两组合（距离满足条件），再次Re-detection，得到更精确的boxes(有相似目标就有几率会有多个box)。将检测到的目标通过关联的方式形成跟踪轨迹链，如果一但有干扰目标存在，那么开辟一条新轨迹，最后得到多条轨迹，根据相邻轨迹之间首尾，尾首目标中心之间的距离来判断是否是同一条轨迹，最终得到目标的最终轨迹。 优点：long-term的测试效果则非常好 缺点：由于re-detection采用级联RCNN，精度高，但速度低 '''

  • PrDiMP: Probabilistic Regression for Visual Tracking. [code]
  • Recursive Least-Squares Estimator-Aided Online Learning for Visual Tracking. [code]
  • ROAM: Recurrently Optimizing Tracking Model. [code]
  • One-Shot Adversarial Attacks on Visual Tracking With Dual Attention.
  • AutoTrack: Towards High-Performance Visual Tracking for UAV With Automatic Spatio-Temporal Regularization. [code]
  • High-Performance Long-Term Tracking With Meta-Updater. [code] 长时跟踪
  • Cooling-Shrinking Attack: Blinding the Tracker with Imperceptible Noises. [code] 目标跟踪鲁棒性研究
  • MAST: A Memory-Augmented Self-Supervised Tracker. [code]

ECCV

  • Learning Feature Embeddings for Discriminant Model based Tracking. [paper]
  • CLNet: A Compact Latent Network for Fast Adjusting Siamese Tracker.
  • Ocean: Learning Object-aware Anchor-free Networks for Real-time Object Tracking. [code]
  • PG-Net: Pixel to Global Matching Network for Visual Tracking.
  • Know Your Surroundings: Exploiting Scene Information for Object Tracking.
  • SPARK: Spatial-aware Online Incremental Attack Against Visual Tracking.
  • Efficient Adversarial Attacks for Visual Object Tracking.

AAAI

  • Discriminative and Robust Online Learning for Siamese Visual Tracking.
  • Exploiting Spatial Invariance for Scalable Unsupervised Object Tracking.
  • GlobalTrack: A Simple and Strong Baseline for Long-term Tracking.
  • SiamFC++: Towards Robust and Accurate Visual Tracking with Target Estimation Guidelines
  • Real-Time Object Tracking via Meta-Learning: Efficient Model Adaptation and One-Shot Channel Pruning
  • SPSTracker: Sub-Peak Suppression of Response Map for Robust Object Tracking

2019年

CVPR

  • Unsupervised Deep Tracking. [code]
  • Target-Aware Deep Tracking. [code]
  • SPM-Tracker: Series-Parallel Matching for Real-Time Visual Object Tracking.
  • SiamRPN++: Evolution of Siamese Visual Tracking With Very Deep Networks. [code]
  • SiamDW: Deeper and Wider Siamese Networks for Real-Time Visual Tracking. [code]
  • GCT:Graph Convolutional Tracking.
  • ATOM: Accurate Tracking by Overlap MaXimization. [code]
  • C-RPN: Siamese Cascaded Region Proposal Networks for Real-Time Visual Tracking. [paper][code]

ICCV

  • DiMP: Learning Discriminative Model Prediction for Tracking. [paper] [code]
  • GradNet: Gradient-Guided Network for Visual Object Tracking. [code]
  • UpdateNet: Learning the Model Update for Siamese Trackers. [[code](https://github.com/zhanglichao/ updatenet)]
  • MLT: Deep Meta Learning for Real-Time Target-Aware Visual Tracking.
  • Joint Group Feature Selection and Discriminative Filter Learning for Robust Visual Object Tracking. [[code](https://github.com/XU-TIANYANG/ GFS-DCF)]
  • SPLT: Skimming-Perusal' Tracking: A Framework for Real-Time and Robust Long-Term Tracking. [code]
  • Learning Aberrance Repressed Correlation Filters for Real-Time UAV Tracking.
  • Bridging the Gap Between Detection and Tracking: A Unified Approach.