[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-72292":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":23,"hasPages":25,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":37,"readmeContent":38,"aiSummary":39,"trendingCount":16,"starSnapshotCount":16,"syncStatus":40,"lastSyncTime":41,"discoverSource":42},72292,"trackers","roboflow\u002Ftrackers","roboflow","Trackers gives you clean, modular re-implementations of leading multi-object tracking algorithms released under the permissive Apache 2.0 license. You combine them with any detection model you already use.","https:\u002F\u002Ftrackers.roboflow.com\u002F",null,"Python",3484,360,35,5,0,23,40,84,69,108.07,"Apache License 2.0",false,"develop",true,[27,28,29,30,31,32,33,34,35,36],"bot-sort","bytetrack","dancetrack","mot17","multi-object-tracking","oc-sort","soccernet","sort","sportsmot","trackeval","2026-06-12 04:01:04","\u003Cdiv align=\"center\">\n    \u003Cimg width=\"200\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Froboflow\u002Ftrackers\u002Frefs\u002Fheads\u002Frelease\u002Fstable\u002Fdocs\u002Fassets\u002Flogo-trackers-violet.svg\" alt=\"trackers logo\">\n    \u003Ch1>trackers\u003C\u002Fh1>\n    \u003Cp>Plug-and-play multi-object tracking for any detection model.\u003C\u002Fp>\n\n[![version](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftrackers.svg)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftrackers)\n[![downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Ftrackers)](https:\u002F\u002Fpypistats.org\u002Fpackages\u002Ftrackers)\n[![license](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flicense-Apache%202.0-blue)](https:\u002F\u002Fgithub.com\u002Froboflow\u002Ftrackers\u002Fblob\u002Frelease\u002Fstable\u002FLICENSE.md)\n[![python-version](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fpyversions\u002Ftrackers)](https:\u002F\u002Fbadge.fury.io\u002Fpy\u002Ftrackers)\n[![colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Froboflow-ai\u002Fnotebooks\u002Fblob\u002Fmain\u002Fnotebooks\u002Fhow-to-track-objects-with-bytetrack-tracker.ipynb)\n[![discord](https:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F1159501506232451173?logo=discord&label=discord&labelColor=fff&color=5865f2&link=https%3A%2F%2Fdiscord.gg%2FGbfgXGJ8Bk)](https:\u002F\u002Fdiscord.gg\u002FGbfgXGJ8Bk)\n\n\u003C\u002Fdiv>\n\nKeeping track of objects across video frames is one of those problems that sounds simple until you try it — occlusions, fast motion, similar-looking targets, and moving cameras all conspire against you. `trackers` gives you clean, benchmarked implementations of SORT, ByteTrack, OC-SORT, and BoT-SORT so you can skip the plumbing and focus on your application. It speaks `supervision.Detections` natively, which means it slots into any detector you already use — YOLO, DETR, RT-DETR, or anything else — without glue code. Whether you are a researcher comparing algorithms, an engineer shipping a production pipeline, or a hobbyist building something cool, `trackers` gives you a single consistent interface for all of them. Requires Python ≥ 3.10.\n\n## Why trackers?\n\n- **Clean-room implementations.** Every algorithm is re-implemented from the original paper — not a thin wrapper around someone else's code. You can read it, understand it, and modify it.\n- **Detector-agnostic.** Works with YOLO, DETR, RT-DETR, or any model that produces bounding boxes. No inference library required or assumed.\n- **`supervision.Detections` native.** Plugs directly into the supervision ecosystem. Pass detections in, get tracked detections back — zero glue code.\n- **Benchmarked across four datasets.** MOT17, SportsMOT, SoccerNet, and DanceTrack — at default parameters and after hyperparameter tuning, so you know what to expect before you deploy.\n- **Tunable out of the box.** Built-in Optuna-based hyperparameter search via `trackers tune` so you can optimize for your specific scene and detector.\n- **Camera motion compensation.** BoT-SORT handles moving cameras natively, keeping track IDs stable even when the whole frame shifts.\n\n## Install\n\n```bash\npip install trackers\n```\n\n\u003Cdetails>\n\u003Csummary>Install from source\u003C\u002Fsummary>\n\n```bash\npip install git+https:\u002F\u002Fgithub.com\u002Froboflow\u002Ftrackers.git\n```\n\n\u003C\u002Fdetails>\n\nFor more options, see the [install guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Finstall\u002F).\n\n[![Watch: Building Real-Time Multi-Object Tracking with RF-DETR and Trackers](https:\u002F\u002Fstorage.googleapis.com\u002Fcom-roboflow-marketing\u002Ftrackers\u002Fdocs\u002Froboflow-piotr-rf-detr-trackers-v1b-callout.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=u0k2dTZ0vfs)\n\n## Quick Start\n\nAdd tracking to your existing detection pipeline in a few lines. Every tracker shares the same `update(detections, frame=None)` interface, so switching algorithms later is a one-line change. The example below uses `inference` as the detector — swap it for any detector that returns `supervision.Detections`.\n\n```python\nimport cv2\nimport supervision as sv\nfrom inference import get_model\nfrom trackers import ByteTrackTracker\n\nmodel = get_model(model_id=\"rfdetr-medium\")\ntracker = ByteTrackTracker()\n\ncap = cv2.VideoCapture(\"video.mp4\")\nwhile cap.isOpened():\n    ret, frame = cap.read()\n    if not ret:\n        break\n\n    result = model.infer(frame)[0]\n    detections = sv.Detections.from_inference(result)\n    tracked = tracker.update(detections)\n```\n\nFor more examples, see the [tracking guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Ftrack\u002F).\n\n## Track from CLI\n\nPrefer the terminal? Point `trackers track` at a video, webcam feed, RTSP stream, or image directory and it handles detection, tracking, and annotated output in one command — no Python script required.\n\n```bash\ntrackers track \\\n    --source video.mp4 \\\n    --output output.mp4 \\\n    --model rfdetr-medium \\\n    --tracker bytetrack \\\n    --show-labels \\\n    --show-trajectories\n```\n\nFor all CLI options, see the [tracking guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Ftrack\u002F).\n\n## Algorithms\n\nEach tracker below is a faithful implementation of its original paper. Pick the one that fits your scene, or run the benchmark to find out which performs best on your data.\n\n|                   Algorithm                   |                           Description                           | MOT17 HOTA | SportsMOT HOTA | SoccerNet HOTA | DanceTrack HOTA |\n| :-------------------------------------------: | :-------------------------------------------------------------: | :--------: | :------------: | :------------: | :-------------: |\n|   [SORT](https:\u002F\u002Farxiv.org\u002Fabs\u002F1602.00763)    |          Kalman filter + Hungarian matching baseline.           |    58.4    |      70.9      |      81.6      |      45.0       |\n| [ByteTrack](https:\u002F\u002Farxiv.org\u002Fabs\u002F2110.06864) | Two-stage association using high and low confidence detections. |    60.1    |      73.0      |      84.0      |      50.2       |\n|  [OC-SORT](https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.14360)  |          Observation-centric recovery for lost tracks.          |    61.9    |      71.7      |      78.4      |    **51.8**     |\n| [BoT-SORT](https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.14651)  |                   Camera motion compensation                    |  **63.7**  |    **73.8**    |    **84.5**    |      50.5       |\n\nAll scores use default parameters on the standard split. See the [tracker comparison](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Ftrackers\u002Fcomparison\u002F) for tuned numbers and methodology.\n\n## Evaluate\n\nOnce you have tracking results, you want to know how good they are. `trackers eval` computes CLEAR, HOTA, and Identity metrics against ground-truth annotations and prints a per-sequence breakdown alongside the combined score.\n\n```bash\ntrackers eval \\\n    --gt-dir .\u002Fdata\u002Fmot17\u002Fval \\\n    --tracker-dir results \\\n    --metrics CLEAR HOTA Identity \\\n    --columns MOTA HOTA IDF1\n```\n\n```\nSequence                        MOTA    HOTA    IDF1\n----------------------------------------------------\nMOT17-02-FRCNN                30.192  35.475  38.515\nMOT17-04-FRCNN                48.912  55.096  61.854\nMOT17-05-FRCNN                52.755  45.515  55.705\nMOT17-09-FRCNN                51.441  50.108  57.038\nMOT17-10-FRCNN                51.832  49.648  55.797\nMOT17-11-FRCNN                55.501  49.401  55.061\nMOT17-13-FRCNN                60.488  58.651  69.884\n----------------------------------------------------\nCOMBINED                      47.406  50.355  56.600\n```\n\nFor the full evaluation workflow, see the [evaluation guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Fevaluate\u002F).\n\n## Download Datasets\n\nNeed benchmark data to evaluate against? `trackers download` pulls MOT17, SportsMOT, and other supported datasets with a single command, handling splits and assets selectively so you only download what you need.\n\n```bash\ntrackers download mot17 \\\n    --split val \\\n    --asset annotations,detections\n```\n\n|   Dataset   |                               Description                               |         Splits         |                Assets                 |     License     |\n| :---------: | :---------------------------------------------------------------------: | :--------------------: | :-----------------------------------: | :-------------: |\n|   `mot17`   |    Pedestrian tracking with crowded scenes and frequent occlusions.     | `train`, `val`, `test` | `frames`, `annotations`, `detections` | CC BY-NC-SA 3.0 |\n| `sportsmot` | Sports broadcast tracking with fast motion and similar-looking targets. | `train`, `val`, `test` |        `frames`, `annotations`        |    CC BY 4.0    |\n\nFor more download options, see the [download guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Fdownload\u002F).\n\n## Try It\n\nWant to see it in action before writing any code? Try trackers in your browser with our [Hugging Face Playground](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Froboflow\u002Ftrackers) — no install required.\n\n## Where to go next\n\n- **New to tracking?** Start with the [tracking guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Ftrack\u002F) — it walks through the Python API and CLI end to end.\n- **Want benchmarks?** The [tracker comparison](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Ftrackers\u002Fcomparison\u002F) covers all four algorithms across all four datasets, at default and tuned parameters, with guidance on which to pick for your scene.\n- **Building a research pipeline?** The [evaluation guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Fevaluate\u002F) and [download guide](https:\u002F\u002Ftrackers.roboflow.com\u002Fdevelop\u002Flearn\u002Fdownload\u002F) cover the full offline benchmarking workflow.\n- **Full API reference** → [trackers.roboflow.com](https:\u002F\u002Ftrackers.roboflow.com)\n- **Try without installing** → [Hugging Face Playground](https:\u002F\u002Fhuggingface.co\u002Fspaces\u002Froboflow\u002Ftrackers)\n- **Questions?** Find us on [Discord](https:\u002F\u002Fdiscord.gg\u002FGbfgXGJ8Bk).\n\n## Contributing\n\nWe welcome contributions. Read our [contributor guidelines](https:\u002F\u002Fgithub.com\u002Froboflow\u002Ftrackers\u002Fblob\u002Frelease\u002Fstable\u002FCONTRIBUTING.md) to get started.\n\n## License\n\nThe code is released under the [Apache 2.0 license](https:\u002F\u002Fgithub.com\u002Froboflow\u002Ftrackers\u002Fblob\u002Frelease\u002Fstable\u002FLICENSE).\n","`trackers` 是一个用于多目标跟踪的Python库，它提供了多种主流跟踪算法的干净、模块化实现，并且可以与任何现有的检测模型无缝集成。项目支持包括SORT、ByteTrack、OC-SORT和BoT-SORT在内的多种算法，每种算法都是基于原始论文重新实现的，确保了代码的可读性和可修改性。此外，该库对检测器没有特定要求，无论是YOLO、DETR还是其他产生边界框的模型都能直接使用。通过提供统一的接口，`trackers`简化了从研究到生产的整个过程，适用于需要在视频序列中持续追踪对象的各种场景，如体育赛事分析、安防监控等。",2,"2026-06-11 03:41:12","high_star"]