visual_tracking

List for public implementation of various algorithms

1) Pubilc Datasets and Challenges

• Human Action Video Dataset
• Awesome-Video-Datasets [Video Dataset Overview]
• VOT Challenge [VOT git repos]
• TAO: A Large-Scale Benchmark for Tracking Any Object (ECCV2020)
• MOT challenge: Multiple Object Tracking Benchmark!

2) Pioneers and Experts

[👍Martin Danelljan] [VIS Lab People]

3) Blogs and Videos

• (B站) 2019-2020年目标跟踪资源全汇总（论文、模型代码、优秀实验室）
• (B站) 张志鹏：Ocean/Ocean+：实时目标跟踪分割算法，小代价，大增益
• (github) Visual Tracking Paper List
• (github) PyQt5_YoLoV5_DeepSort
• (github) [YOLOv5 + DeepSORT] Yolov5 deepsort inference，使用YOLOv5+Deepsort实现车辆行人追踪和计数
• (github) [YOLOv5 + DeepSORT] Real-time multi-object tracker using YOLO v5 and deep sort[Evaluation Page]
• (github) [Official DeepSort] MOT tracking using deepsort and yolov3 with pytorch
• (github) [YOLOv5 + SORT] ClassySORT is a simple real-time multi-object tracker (MOT)
• (github) [FastMOT] High-performance multiple object tracking based on YOLO, Deep SORT, and KLT
• (github} [IJCV-2021] FairMOT: On the Fairness of Detection and Re-Identification in Multi-Object Tracking
• (arxiv) CountingMOT: Joint Counting, Detection and Re-Identification for Multiple Object Tracking

4) Papers and Sources Codes

▶ Survey

▶ Optical Flow

• FlowNet2.0(CVPR2017) FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks [arxiv link][Codes|PyTorch(offical)]

▶ Siamese Based

• SiamMask(CVPR2019) Fast Online Object Tracking and Segmentation: A Unifying Approach [arxiv link][project link][Codes|PyTorch(offical)][blog in wechat][blog in bilibili]

▶ CF Based

• ECO(CVPR2017) ECO: Efficient Convolution Operators for Tracking [arxiv link][Codes|Matlab(offical)][Codes|MXNet(unoffical)][CSDN blog]

▶ Subfield

• SiamCAR(CVPR2020) SiamCAR: Siamese Fully Convolutional Classification and Regression for Visual Tracking [arxiv link][cvpr link][Codes|offical PyTorch]

▶ Point Based

Point tracking is a new field with a few notable works released around the same time.

• PIPs(ECCV2022 Oral)(arxiv2022.04) Particle Video Revisited: Tracking Through Occlusions Using Point Trajectories [paper link][arxiv link][project link][code|official][Carnegie Mellon University][It was an inspiration for the work TAPIR]

• 👍OmniMotion(ICCV2023 Oral, Best Student Paper)(arxiv2023.06) Tracking Everything Everywhere All at Once [paper link][arxiv link][project link][code|official][Cornell University + Google Research + UC Berkeley][It doesn't perform as well as TAPIR and is substantially slower, but it provides pseudo-3D reconstructions, and could potentially be used on top of TAPIR tracks to further improve performance.]

• TAPIR(ICCV2023)(arxiv2023.06) TAPIR: Tracking Any Point with per-frame Initialization and temporal Refinement [paper link][arxiv link][project link][code|official][Google DeepMind + University of Oxford]

• MFT(WACV2024)(arxiv2023.05) MFT: Long-Term Tracking of Every Pixel [paper link][arxiv link][project link][code|official][Czech Technical University in Prague][Multi-Flow Tracking hypothesizes many flow fields between different pairs of frames and scores them; multiple hypotheses leads to improved robustness.]

• LEAP-VO(CVPR2024)(arxiv2024.01) LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry [paper link][arxiv link][project link][code|official][TU Munich + Munich Center for Machine Learning + MPI for Intelligent Systems + Microsoft][Tracking + Visual Odometry, Long-term Effective Any Point Tracking (LEAP)]

• 👍CoTracker(ECCV2024)(arxiv2023.07) CoTracker: It is Better to Track Together [paper link][arxiv link][project link][code|official][Meta AI + Visual Geometry Group, University of Oxford]

• 👍DINO-Tracker(ECCV2024)(arxiv2024.03) DINO-Tracker: Taming DINO for Self-Supervised Point Tracking in a Single Video [paper link][arxiv link][project link][code|official][Weizmann Institute of Science, Israel]

• BootsTAP(arxiv2024.02) BootsTAP: Bootstrapped Training for Tracking-Any-Point [arxiv link][project link][code|official][Google DeepMind + University of Oxford][It is based on the TAPIR]

• **** [paper link][arxiv link][project link][code|official]