{"full_name":"ultralytics/ultralytics","owner":"ultralytics","name":"ultralytics","shard":"1","url":"https://github.com/ultralytics/ultralytics","description":"Ultralytics YOLO27, YOLO26, YOLO11, YOLOv8 — object detection, instance segmentation, semantic segmentation, image classification, pose estimation, object tracking","homepage":"https://platform.ultralytics.com","topics":["ultralytics","yolov8","yolo","deep-learning","image-classification","machine-learning","object-detection","pytorch","tracking","instance-segmentation","yolo11","computer-vision","python","pose-estimation","rotated-object-detection"],"tag_pages":["computer-vision","deep-learning","image-classification","instance-segmentation","machine-learning","object-detection","pose-estimation","python","pytorch","tracking","yolo"],"language":"Python","license":"AGPL-3.0","avatar":"https://avatars.githubusercontent.com/u/26833451?v=4","created_at":"2022-09-11T16:39:45Z","pushed_at":"2026-11-15T19:25:04Z","metrics":{"stars":63456,"forks":12012,"watchers":265,"issues":13003,"prs":8414,"releases":540,"issues_open":45,"issues_closed":12958,"prs_open":29,"prs_closed":1970,"prs_merged":6415,"discussions":2237},"growth":{"day":{"stars":51,"forks":10,"base_day":"2026-11-15"},"week":{"stars":226,"forks":43,"base_day":"2026-11-09"}},"history":{"dates":["2026-11-09","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[63230,63266,63300,63323,63328,63356,63405,63456],"forks":[11969,11973,11977,11983,11986,11986,12002,12012],"watchers":[null,null,266,266,266,265,265,265],"issues_open":[null,null,47,47,47,45,45,45],"prs_open":[null,null,35,35,35,29,29,29]},"boards":{},"same_owner":[{"full_name":"ultralytics/yolov5","name":"yolov5","shard":"1","boards":{}}],"issues":[{"number":1429,"title":"NEW - YOLOv8 🚀 Multi-Object Tracking","comments":175,"created_at":"2023-03-14T20:39:33Z"},{"number":1915,"title":"NEW - YOLOv8 🚀 Pose Models","comments":106,"created_at":"2023-04-09T16:55:09Z"},{"number":16602,"title":"Ultralytics YOLO11 Launch 🚀","comments":46,"created_at":"2024-10-01T16:32:40Z"},{"number":21479,"title":"Allowing Multiple Reference Images for YOLOE (Again)","comments":30,"created_at":"2025-07-21T21:09:18Z"},{"number":6921,"title":"Add new solution - action recognition","comments":23,"created_at":"2023-12-11T14:45:22Z"},{"number":6168,"title":"Custom callback functions doesn't called in DDP training","comments":21,"created_at":"2023-11-07T06:09:17Z"},{"number":25695,"title":"YOLO-Depth: metric per-pixel distance from a single camera","comments":13,"created_at":"2026-08-05T14:08:45Z"},{"number":26324,"title":"Extend inference prefetch beyond CUDA: MPS, accelerator backends, CPU runtimes and videos","comments":11,"created_at":"2026-09-24T18:33:33Z"}],"discussions":[{"number":27149,"title":"Overcoming False Positives in Real-Time Computer Vision for Engine Room Oil Leak Detection","comments":4,"category":"Q&A","created_at":"2026-11-11T06:20:37Z"},{"number":27107,"title":"difference between Object Detection and Object Segmentation","comments":2,"category":"Q&A","created_at":"2026-11-08T15:21:09Z"},{"number":27106,"title":"file format","comments":2,"category":"Q&A","created_at":"2026-11-08T15:15:57Z"},{"number":27105,"title":"file format","comments":2,"category":"Docs","created_at":"2026-11-08T15:13:00Z"},{"number":27036,"title":"Content Under Review","comments":0,"category":"Show and tell","created_at":"2026-11-04T16:27:24Z"},{"number":26929,"title":"guides/raspberry-pi/","comments":2,"category":"Docs","created_at":"2026-10-28T16:25:41Z"},{"number":26911,"title":"Confusing logic in TAL assigner: `count_tensor.masked_fill_(count_tensor > 1, 0)` + subsequent `select_highest_overlaps`","comments":2,"category":"Ideas","created_at":"2026-10-27T09:52:13Z"},{"number":26909,"title":"Clarification on negative bounding boxes in the Brain Tumor dataset","comments":2,"category":"Q&A","created_at":"2026-10-27T07:49:34Z"}],"latest_release":{"name":"v8.4.212 - `ultralytics 8.4.212` Read Korean, Cyrillic, Arabic and other scripts with OCR (#27167)","tag":"v8.4.212","published_at":"2026-11-13T13:08:32Z","body":"## 🌟 Summary\n**Ultralytics 8.4.212 expands OCR to more writing systems and improves Platform discovery, data import, training reliability, and inference compatibility.**\n\n## 📊 Key Changes\n- 🌍 **OCR supports more languages and scripts.** Choose a PP-OCRv5 model to recognize Korean, Cyrillic, Arabic, Devanagari, Thai, Greek, Tamil, or Telugu text. These models run through ONNX Runtime, without requiring PaddlePaddle or PaddleOCR. For example:\n  ```python\n  ocr = solutions.OCR(recognizer=\"korean_PP-OCRv5_mobile_rec\")\n  result = ocr(im0)\n  ```\n  Select the model for the text you expect; OCR does not automatically identify or switch languages. Hebrew is not included in the new models.\n- 🔎 **Platform Explore adds a Models tab.** Search public models by what they detect, browse results, and try or clone models. The Platform docs also cover direct imports from hosted Label Studio, including supported detection, oriented-box, segmentation, and classification projects.\n- 🧰 **Training and evaluation are more dependable.** Fixes prevent distributed training from hanging when canceled, restore deterministic training for certain tasks on older macOS MPS setups, and include mask, pose, and semantic pixel-accuracy details in exported validation summaries.\n- ⚡ **Inference and export fixes improve compatibility and efficiency.** ONNX Runtime is better matched to CUDA 12 PyTorch installs and uses more suitable CPU thread limits; TensorRT checks recognize installations in NVIDIA containers and JetPack. MNN inputs, YOLOE tracking with ReID, and Edge TPU pose keypoints also receive fixes.\n- 🛡️ **Data handling is safer and more efficient.** Detection crop saving avoids unnecessary full-image copies; segmentation-mask conversion now rejects filename collisions before overwriting labels; source lists consistently use UTF-8.\n- 📚 **Documentation is refreshed.** Guides reflect current Platform features, and outdated DeepSparse documentation is removed because the project has reached end of life.\n\n## 🎯 Purpose & Impact\n- People working with multilingual images can use a suitable OCR recognizer for scripts that the default PP-OCRv6 model does not cover.\n- Platform users can more easily find relevant public models and bring supported hosted Label Studio projects into their workflow.\n- Training, validation exports, and deployment should behave more reliably across distributed setups and a wider range of hardware and software versions.\n- Safer file handling and reduced image copying help prevent accidental data loss and avoid unnecessary work.\n\n## What's Changed\n* Platform docs: Explore Models tab, HEIC previews in Predict, Recent Activity dataset status by @pderrenger in https://github.com/ultralytics/ultralytics/pull/27160\n* Platform docs: current VLM leaderboard picks, Explore model deploy and ranking wording, Predict screenshots without Ask AI by @glenn-jocher in https://github.com/ultralytics/ultralytics/pull/27163\n* Fix DDP hang when Platform cancels training mid-epoch by @artest08 in https://github.com/ultralytics/ultralytics/pull/27161\n* Fix deterministic semantic, depth and segment training on MPS with torch<2.3 by @raimbekovm in https://github.com/ultralytics/ultralytics/pull/27159\n* Divide the dataloader worker cap by the GPUs of the run by @raimbekovm in https://github.com/ultralytics/ultralytics/pull/27162\n* Update YouTube video link in `isolating-segmentation-objects` docs by @RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/27165\n* Remove neural magic deepsparse docs pages after end of life by @raimbekovm in https://github.com/ultralytics/ultralytics/pull/27169\n* Avoid full-frame copies when saving detection crops by @aswanth-07 in https://github.com/ultralytics/ultralytics/pull/27170\n* Fix YOLOE tracking with native ReID features by @raimbekovm in https://github.com/ultralytics/ultralytics/pull/27171\n* Read source manifests as UTF-8 by @mh-masum6875 in https://github.com/ultralytics/ultralytics/pull/27172\n* Prevent mask convers","is_prerelease":false},"readme":true,"readme_zh":true,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}