Ultralytics v8.2.0 Release Notes

Introduction

Ultralytics is excited to announce the v8.2.0 release of YOLOv8, comprising 277 merged Pull Requests by 32 contributors since our last v8.1.0 release in January 2024, marking another milestone in our journey to make state-of-the-art AI accessible and powerful. This release brings a host of new features, performance optimizations, and expanded integrations, reflecting our commitment to continuous improvement and innovation. 🌍🚀

Ultralytics v8.2.0 Key Highlights

• New Models: Introduced support for YOLOv8-World, YOLOv8-World-v2 (by @Laughing-q in PR #9268), YOLOv9-C, YOLOv9-E (by @Laughing-q in PR #8571), and YOLOv9 Segment models (by @Burhan-Q in PR #9296), expanding the versatility of the Ultralytics platform.
• New Features: Added distance calculation in vision-eye, per-class object counting (by @RizwanMunawar in PR #9443), and queue management utilities (by @RizwanMunawar in PR #9494), enhancing the functionality and applicability of YOLOv8.
• Performance Optimizations: Achieved 40% faster ultralytics imports (by @glenn-jocher in PR #9547), faster batch same_shapes, and immediate checkpoint serialization (by @glenn-jocher in PR #9437), further optimizing the efficiency of the framework.
• Enhanced Export Capabilities: Improved export support, including OpenVINO 2023.3 updates (by @adrianboguszewski in PR #8417), TensorRT 10 support (by @Burhan-Q in PR #9516), and fixes for TFLite, ONNX, and OpenVINO exports.
• Documentation Expansion: Significantly expanded the documentation with new guides, integration pages for TorchScript, TFLite, NCNN, PaddlePaddle, TF GraphDef, TF SavedModel, TF.js (by @abirami-vina in multiple PRs), and updates to existing pages, providing comprehensive resources for users.
• Training Enhancements: Introduced YOLO-World training support (by @Laughing-q in PR #9268), fixed learning rate issues (by @Laughing-q in PR #9468), and improved robustness for stopping and resuming training (by @glenn-jocher in PR #9384).
• Platform Support: Added support for NVIDIA Jetson (by @lakshanthad in PR #9484), Raspberry Pi (by @lakshanthad in PR #8828), and Apple M1 runners for tests and benchmarks (by @glenn-jocher in PR #8162), expanding the usability of YOLOv8 across various platforms.
• CI/CD Improvements: Enhanced Ultralytics Actions using OpenAI GPT-4 for PR summaries (by @pderrenger in PR #7867) and introduced self-hosted Raspberry Pi 5 CI (by @lakshanthad in PR #8828), streamlining the development and testing processes.
• Bug Fixes: Resolved various issues related to model loading, inference, plotting, and exports, ensuring a smoother user experience.
• Community Contributions: Welcomed contributions from 31 new contributors, reflecting the growing engagement and collaborative spirit within the Ultralytics community.

Summary

Ultralytics v8.2.0 represents a significant leap forward, introducing new models, features, and optimizations while expanding platform support and integration capabilities. We extend our gratitude to our dedicated users and contributors for their invaluable support and contributions. As we continue to push the boundaries of AI and computer vision, we look forward to the exciting possibilities and advancements that lie ahead! 🌟🚀🎉

What's Changed

• YOLOv8.1 blog, Explorer notebook and 2023 > 2024 updates by @AyushExel in ultralytics/ultralytics#7469
• Explorer with LanceDB, Actions and Docs updates by @glenn-jocher in ultralytics/ultralytics#7487
• OBB Docs updates by @glenn-jocher in ultralytics/ultralytics#7512
• Update OpenVINO INT8 export by @glenn-jocher in ultralytics/ultralytics#7515
• ultralytics 8.1.1 Docs, Solutions and Export updates by @glenn-jocher in ultralytics/ultralytics#7545
• Update HTTP to HTTPS by @glenn-jocher in ultralytics/ultralytics#7548
• Python refactorings and simplifications by @glenn-jocher in ultralytics/ultralytics#7549
• Use pathlib in DOTA ops by @glenn-jocher in ultralytics/ultralytics#7552
• OBB Docs updates by @glenn-jocher in ultralytics/ultralytics#7568
• Update YOLOv3 and YOLOv5 YAMLs by @glenn-jocher in ultralytics/ultralytics#7574
• Add docstrings to new HUB functions by @glenn-jocher in ultralytics/ultralytics#7576
• OBB: Fix plot_images by @Laughing-q in ultralytics/ultralytics#7592
• OBB: update metrics by @Laughing-q in ultralytics/ultralytics#7593
• Resize angle, count, and stage on keypoint number change by @gvzdv in ultralytics/ultralytics#7598
• Mkdocs annotations fixes by @glenn-jocher in ultralytics/ultralytics#7600
• ultralytics 8.1.2 scope HUB-SDK imports by @glenn-jocher in ultralytics/ultralytics#7596
• Update docs building code by @glenn-jocher in ultralytics/ultralytics#7601
• YAML reformat by @glenn-jocher in ultralytics/ultralytics#7669
• Add PR Summary step to Ultralytics Actions by @glenn-jocher in ultralytics/ultralytics#7675
• Fixed dataloader CPU bottleneck for small batch sizes by @ExtReMLapin in ultralytics/ultralytics#7659
• Update mkdocs.yml by @glenn-jocher in ultralytics/ultralytics#7693
• ultralytics 8.1.3 ResNet models and lighter dependencies by @glenn-jocher in ultralytics/ultralytics#7700
• Update Twitter icon in Docs by @glenn-jocher in ultralytics/ultralytics#7711
• ultralytics 8.1.4 RTDETR TensorBoard graph visualization fix by @glenn-jocher in ultralytics/ultralytics#7725
• Update Docs robots.txt by @glenn-jocher in ultralytics/ultralytics#7728
• Bounding Box to OBB conversion by @Burhan-Q in ultralytics/ultralytics#7572
• Add yolo_bbox2segment docs reference by @glenn-jocher in ultralytics/ultralytics#7751
• ultralytics 8.1.5 add OBB Tracking support by @Laughing-q in ultralytics/ultralytics#7731
• Clean up unused imgsz by @Laughing-q in ultralytics/ultralytics#7771
• Add HUB-SDK docs by @glenn-jocher in ultralytics/ultralytics#7775
• Add OBB benchmarks to CI by @glenn-jocher in ultralytics/ultralytics#7777
• Add YOLOv8-OBB https://youtu.be/Z7Z9pHF8wJc by @glenn-jocher in ultralytics/ultralytics#7780
• Update H1 in Explorer API docs by @RizwanMunawar in ultralytics/ultralytics#7813
• Adds toggle displaying labels in GUI and verbose log on start by @AyushExel in ultralytics/ultralytics#7804
• Fix bbox2segment converter by @Laughing-q in ultralytics/ultralytics#7814
• Add ONNX Docs integrations page by @abirami-vina in ultralytics/ultralytics#7802
• Fix Yolo 8.0.206 scale bug by @Alarmod in ultralytics/ultralytics#7821
• ultralytics 8.1.6 revert 8.0.206 box ops box scaling by @glenn-jocher in ultralytics/ultralytics#7823
• Explorer API video https://youtu.be/3VryynorQeo by @RizwanMunawar in ultralytics/ultralytics#7838
• Add HUB-SDK Docs reference section by @glenn-jocher in ultralytics/ultralytics#7781
• Link checks SSL insecure robustness by @glenn-jocher in ultralytics/ultralytics#7853
• Add new @Retry() decorator by @glenn-jocher in ultralytics/ultralytics#7854
• Add TensorRT Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#7855
• Cleanup Docs languages by @glenn-jocher in ultralytics/ultralytics#7865
• Add millimeters in solutions/distance_caculation.py + object-cropping.md visuals by @RizwanMunawar in ultralytics/ultralytics#7860
• ultralytics 8.1.7 USER_CONFIG_DIR Explorer ops by @AyushExel in ultralytics/ultralytics#7861
• Ultralytics Actions with OpenAI GPT-4 PR Summary by @pderrenger in ultralytics/ultralytics#7867
• Bump slackapi/slack-github-action from 1.24.0 to 1.25.0 in /.github/workflows by @dependabot in ultralytics/ultralytics#7871
• Add visuals in guides/distance-calculation.md by @RizwanMunawar in ultralytics/ultralytics#7876
• TensorRT FP16 export with dynamic=True by @developer0hye in ultralytics/ultralytics#7870
• Add https://youtu.be/3VryynorQeo to README and fix converter.py Docs by @RizwanMunawar in ultralytics/ultralytics#7883
• ultralytics 8.1.8 new model.save('filename.pt') method by @glenn-jocher in ultralytics/ultralytics#7886
• Compress explorer image in dataset/index.md + description added for new segmentation datasets by @RizwanMunawar in ultralytics/ultralytics#7901
• Fix OpenVINO links by @glenn-jocher in ultralytics/ultralytics#7930
• Add integrations/gradio Docs page by @glenn-jocher in ultralytics/ultralytics#7935
• Add https://youtu.be/96NkhsV-W1U to docs and heatmaps fix by @RizwanMunawar in ultralytics/ultralytics#7944
• Segment TensorRT batch=3 fix by @DieterTimmers in ultralytics/ultralytics#7952
• ultralytics 8.1.9 replace .size(0) with .shape[0] by @glenn-jocher in ultralytics/ultralytics#7957
• Add YouTube iframe loading="lazy" by @glenn-jocher in ultralytics/ultralytics#8001
• Mkdocs updates by @glenn-jocher in ultralytics/ultralytics#8008
• Tracker: clean up GMC by @Laughing-q in ultralytics/ultralytics#8003
• Bump nick-invision/retry from 2 to 3 in /.github/workflows by @dependabot in ultralytics/ultralytics#8023
• Bump codecov/codecov-action from 3 to 4 in /.github/workflows by @dependabot in ultralytics/ultralytics#8022
• Fix TFLite INT8 for OBB by @adamp87 in ultralytics/ultralytics#7989
• Add CoreML Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8063
• Add Roboflow 100 Dataset Docs Page by @abirami-vina in ultralytics/ultralytics#8065
• ultralytics 8.1.10 MLFlow, OBB, TFLite and INT8 fixes by @Youho99 in ultralytics/ultralytics#8016
• Add https://youtu.be/R42s2zFtNIY to hub/datasets.md and CoreML image fix by @RizwanMunawar in ultralytics/ultralytics#8085
• Update Roboflow segmentation Docs pages by @RizwanMunawar in ultralytics/ultralytics#8094
• FROM pytorch/pytorch:2.2.0-cuda12.1-cudnn8-runtime by @glenn-jocher in ultralytics/ultralytics#8104
• ultralytics 8.1.11 expand OpenVINO INT8 ops for improved mAP by @glenn-jocher in ultralytics/ultralytics#7516
• Add https://youtu.be/q7LwPoM7tSQ to guides/yolo-performance-metrics.md and fix cover image for distance_calculation.md by @RizwanMunawar in ultralytics/ultralytics#8114
• Compress docs images by @RizwanMunawar in ultralytics/ultralytics#8124
• Updated SQL Query Plot Code by @0xSynapse in ultralytics/ultralytics#8120
• Elaborate Train and Predict mode docs arguments by @glenn-jocher in ultralytics/ultralytics#8137
• Move Google Drive to slow tests by @glenn-jocher in ultralytics/ultralytics#8163
• Improve Coral Edge TPU guide by @Skillnoob in ultralytics/ultralytics#8160
• Try Apple M1 runners for Tests and Benchmarks CI by @glenn-jocher in ultralytics/ultralytics#8162
• Retry once for Docker CI failures by @glenn-jocher in ultralytics/ultralytics#8164
• ultralytics 8.1.12 new Results.show() and Results.save() by @glenn-jocher in ultralytics/ultralytics#8166
• Fix solution functions description + integrations/openvino.md cover image update by @RizwanMunawar in ultralytics/ultralytics#8170
• Pin duckdb<=0.9.2 to avoid 0.10.0 errors by @AyushExel in ultralytics/ultralytics#8181
• New HUB Cloud Training docs page by @priytosh-tripathi in ultralytics/ultralytics#8174
• ultralytics 8.1.13 fix yolo train time= bug by @glenn-jocher in ultralytics/ultralytics#8179
• Docs updates and improvements by @glenn-jocher in ultralytics/ultralytics#8183
• Fix xyxyxyxyn calculation in obb result by @KybernetikJo in ultralytics/ultralytics#8188
• ultralytics 8.1.14 new YOLOv8-World models by @Laughing-q in ultralytics/ultralytics#8054
• Fix PIL show() and save() by @glenn-jocher in ultralytics/ultralytics#8202
• Remove OpenVINO INT8 RTDETRDecoder ops by @glenn-jocher in ultralytics/ultralytics#8219
• Fix xyxyxyxyn calculation, swap axis by @KybernetikJo in ultralytics/ultralytics#8249
• Improve Docs Modes tables by @glenn-jocher in ultralytics/ultralytics#8266
• ultralytics 8.1.15 add Python 3.12 compatibility by @glenn-jocher in ultralytics/ultralytics#8210
• Expand Model method type hinting by @glenn-jocher in ultralytics/ultralytics#8279
• ultralytics 8.1.16 OBB ConfusionMatrix support by @Laughing-q in ultralytics/ultralytics#8299
• Move Explorer tests to scheduled CI by @glenn-jocher in ultralytics/ultralytics#8305
• Remove JAX constraints in Exporter by @glenn-jocher in ultralytics/ultralytics#8309
• Model typehints Docker fix by @glenn-jocher in ultralytics/ultralytics#8306
• Fix an issue with flatbuffer version for Raspberry Pi by @lakshanthad in ultralytics/ultralytics#8311
• Update Dockerfile for Raspberry Pi Support by @lakshanthad in ultralytics/ultralytics#8123
• Avoid creating -hub dir without processing images by @glenn-jocher in ultralytics/ultralytics#8334
• Added check of parameter type before setting required_grad=True for frozen layers by @alexsu52 in ultralytics/ultralytics#8318
• Add estimate_speed(region_color) parameter by @hi-sushanta in ultralytics/ultralytics#8285
• Add simple-utilities.md Docs page by @Burhan-Q in ultralytics/ultralytics#8269
• Update mkdocs-ultralytics-plugin>=0.0.44 by @glenn-jocher in ultralytics/ultralytics#8347
• ultralytics 8.1.17 fix ClassificationDataset caching by @glenn-jocher in ultralytics/ultralytics#8358
• Add missing single-line docstrings by @glenn-jocher in ultralytics/ultralytics#8362
• ultralytics 8.1.18 add cmake for building onnxsim on aarch64 by @lakshanthad in ultralytics/ultralytics#8395
• Allow Annotator PIL.Image inputs by @glenn-jocher in ultralytics/ultralytics#8397
• Add Non-Maximum Suppression (NMS) inplace flag by @AlainSchoebi in ultralytics/ultralytics#8368
• Compare plt.get_backend() in lowercase by @Mactarvish in ultralytics/ultralytics#8409
• Improve Docs arguments tables by @Burhan-Q in ultralytics/ultralytics#8415
• Add plot_images conf_thresh parameter by @deanmark in ultralytics/ultralytics#8446
• Add YOLOv9 Docs page by @glenn-jocher in ultralytics/ultralytics#8478
• Rename model_id to model.id by @kaikai23 in ultralytics/ultralytics#8447
• ultralytics 8.1.19 PNNX aarch64 linux fix by @glenn-jocher in ultralytics/ultralytics#8480
• Add TorchScript Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8501
• Do RTDETR file suffix check using pathlib instead of string manipulations by @birdcolour in ultralytics/ultralytics#8525
• Add TFLite Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8522
• Fix undefined ‘im_array’ bug in predict.md by @glenn-jocher in ultralytics/ultralytics#8565
• Add https://youtu.be/ie3vLUDNYZo and other YT videos in Docs by @RizwanMunawar in ultralytics/ultralytics#8551
• Hold failed upload metrics and upload with next upload metrics by @hassaanfarooq01 in ultralytics/ultralytics#8513
• ultralytics 8.1.20 add YOLOv8x-World support by @Laughing-q in ultralytics/ultralytics#8539
• Fix device counting method to account for double-digit device IDs by @omarzd in ultralytics/ultralytics#8502
• Remove unused variables by @Kayzwer in ultralytics/ultralytics#8511
• Add NCNN Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8562
• Optimize function calls to method calls by @Kayzwer in ultralytics/ultralytics#8507
• Replacing Images for torchscript.md by @abirami-vina in ultralytics/ultralytics#8596
• Fix cmake requirements for ARM64 by @lakshanthad in ultralytics/ultralytics#8589
• Add https://youtu.be/rCggzXRRSRo to Speed Estimation Docs by @RizwanMunawar in ultralytics/ultralytics#8577
• Add KaTeX support + updates to YOLOv9 model page by @Burhan-Q in ultralytics/ultralytics#8531
• Replace onnx2tf usage from CLI to Python by @ALMerrill in ultralytics/ultralytics#8429
• Update IoU capitalization by @glenn-jocher in ultralytics/ultralytics#8604
• Update TFLite Docs images by @glenn-jocher in ultralytics/ultralytics#8605
• Add distance calculation feature in vision-eye by @glenn-jocher in ultralytics/ultralytics#8616
• ultralytics 8.1.21 Add YOLOv8-World-v2 models by @Laughing-q in ultralytics/ultralytics#8580
• ultralytics 8.1.22 HUB model pathlib fix by @glenn-jocher in ultralytics/ultralytics#8621
• ultralytics 8.1.23 add YOLOv9-C and E models by @Laughing-q in ultralytics/ultralytics#8571
• Add settings runs_dir == datasets_dir warning by @glenn-jocher in ultralytics/ultralytics#8653
• Add Ultralytics HUB Cloud Training banner to Docs by @sergiuwaxmann in ultralytics/ultralytics#8656
• Update CITATION.cff with cffinit corrections by @glenn-jocher in ultralytics/ultralytics#8658
• Update Colab notebook for YOLOv8.1 by @glenn-jocher in ultralytics/ultralytics#8660
• ultralytics 8.1.24 new OpenVINO 2023.3 export updates by @adrianboguszewski in ultralytics/ultralytics#8417
• OBB: Fix when training on zip data by @Laughing-q in ultralytics/ultralytics#8680
• Object Counter improvements by @hdnh2006 in ultralytics/ultralytics#8648
• Clarify triton server boolean value by @bobyard-com in ultralytics/ultralytics#8532
• Add model.eval() in TensorBoad graph visualization to avoid BN stats changes by @Y-T-G in ultralytics/ultralytics#8629
• Add domain checks to CI docs page by @glenn-jocher in ultralytics/ultralytics#8714
• ultralytics 8.1.25 OpenVINO LATENCY and THROUGHPUT modes by @glenn-jocher in ultralytics/ultralytics#8058
• Add https://youtu.be/OpWpBI35A5Y to inference-api Docs by @RizwanMunawar in ultralytics/ultralytics#8717
• OpenVino 2024.0.0 by @johnnynunez in ultralytics/ultralytics#8726
• Fix tfjs exports for ARM64 Dockerfile by @lakshanthad in ultralytics/ultralytics#8766
• Modify tuner best.pt logic to train first by @finlaymorrison in ultralytics/ultralytics#8792
• Update CI.md with CI links by @glenn-jocher in ultralytics/ultralytics#8786
• Add https://youtu.be/ZzUSXQkLbNw to Integration Docs by @RizwanMunawar in ultralytics/ultralytics#8793
• ultralytics 8.1.25 fix **kwargs: (dict) warnings by @glenn-jocher in ultralytics/ultralytics#8815
• Add docs guide terminal images by @Burhan-Q in ultralytics/ultralytics#8819
• RTDETRDetectionModel TorchScript, ONNX Predict and Val support by @glenn-jocher in ultralytics/ultralytics#8818
• New MLFLOW_KEEP_RUN_ACTIVE flag for enhanced MLflow run management by @massimiliano96 in ultralytics/ultralytics#8808
• ultralytics 8.1.26 LoadImagesAndVideos batched inference by @glenn-jocher in ultralytics/ultralytics#8817
• Update fraction arg to employ a random selection by @johnnynunez in ultralytics/ultralytics#8234
• Batch inference fixes by @Burhan-Q in ultralytics/ultralytics#8854
• Integrate OpenVINO CUMULATIVE_THROUGHPUT mode batched inference by @glenn-jocher in ultralytics/ultralytics#8834
• Add PaddlePaddle Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8858
• ultralytics 8.1.27 batched tracking fixes by @Laughing-q in ultralytics/ultralytics#8842
• Faster batch same_shapes by @glenn-jocher in ultralytics/ultralytics#8851
• More robust PNNX downloads by @glenn-jocher in ultralytics/ultralytics#8866
• Fix test_mlflow_keep_run_active() by @glenn-jocher in ultralytics/ultralytics#8868
• Add polygon regions drawing support in object-counting.md and minor docs update by @RizwanMunawar in ultralytics/ultralytics#8885
• Add TFLite Edge TPU Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#8900
• Update Docs README by @glenn-jocher in ultralytics/ultralytics#8919
• Fix names bug when exporting YOLOv8-World to ONNX by @jamjamjon in ultralytics/ultralytics#8941
• New test for labels and crops by @Burhan-Q in ultralytics/ultralytics#8861
• Optimize TFJS export on ARM64 by @lakshanthad in ultralytics/ultralytics#8946
• ultralytics 8.1.28 avoid * ops on bool Tensors for RT-DETR OpenVINO export by @Kayzwer in ultralytics/ultralytics#8937
• Add https://youtu.be/YDXKa1EljmU and https://youtu.be/5BO0Il_YYAg to Docs by @RizwanMunawar in ultralytics/ultralytics#8958
• ultralytics 8.1.29 improved disk space checking on correct path by @glenn-jocher in ultralytics/ultralytics#8977
• Add OpenVINO Latency vs Throughput modes Docs page by @glenn-jocher in ultralytics/ultralytics#9001
• Add brain-tumor dataset docs page by @RizwanMunawar in ultralytics/ultralytics#9041
• Fix Neural Magic links by @glenn-jocher in ultralytics/ultralytics#9144
• Updating a typo on paddlepaddle.md by @abirami-vina in ultralytics/ultralytics#9108
• ultralytics 8.1.30 add advanced HUB train arguments by @kalenmike in ultralytics/ultralytics#9110
• fix yolo classify model loading error by @DseidLi in ultralytics/ultralytics#9196
• Add TF GraphDef Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#9203
• ultralytics 8.1.31 NCNN and CLIP updates by @glenn-jocher in ultralytics/ultralytics#9235
• Fix TypeError in check_cfg() by @zhaoruibing in ultralytics/ultralytics#9245
• Add https://youtu.be/aeAX6vWpfR0 to docs & african-wildlife dataset docs page by @RizwanMunawar in ultralytics/ultralytics#9102
• Add TF SavedModel Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#9162
• Limit ray<2.9.3 due to bug in ray==2.10.0 by @glenn-jocher in ultralytics/ultralytics#9254
• ultralytics 8.1.32 fix CLIP backwards compatibility by @glenn-jocher in ultralytics/ultralytics#9253
• Conda cv2 ImportError libEGL.so.1 fix by @glenn-jocher in ultralytics/ultralytics#9255
• Fix IS_PYTHON_3_12 bug by @glenn-jocher in ultralytics/ultralytics#9258
• Reverse IoU threshold for NMS docs descriptions by @plashchynski in ultralytics/ultralytics#9151
• Add bgr hyperparameter by @Laughing-q in ultralytics/ultralytics#9139
• Update pyproject.toml [tool.pytest.ini_options] by @glenn-jocher in ultralytics/ultralytics#9259
• Fix OpenVINO missing batch-size bug by @glenn-jocher in ultralytics/ultralytics#9264
• ultralytics 8.1.33 fix HUB model checks by @kalenmike in ultralytics/ultralytics#9153
• Update Discord and Contributing Guide URLs by @glenn-jocher in ultralytics/ultralytics#9270
• Add https://youtu.be/uDrn9QZJ2lk and object_counting.py, ai_gym.py updates by @RizwanMunawar in ultralytics/ultralytics#9265
• ultralytics 8.1.34 Inference API robust imgsz checks by @glenn-jocher in ultralytics/ultralytics#9274
• Bump contributor-assistant/github-action from 2.3.1 to 2.3.2 in /.github/workflows by @dependabot in ultralytics/ultralytics#9279
• Add https://youtu.be/u3pLlgzUeV8 to docs by @RizwanMunawar in ultralytics/ultralytics#9367
• Fix typo in YOLOv8-Libtorch-CPP-Inference by @higor-melo in ultralytics/ultralytics#9330
• Simplify metrics calculation by @Kayzwer in ultralytics/ultralytics#9338
• Fix typo in example readme by @ericphanson in ultralytics/ultralytics#9306
• Specify YOLODataset task arg during INT8 calibration by @sifan-intel in ultralytics/ultralytics#9309
• ultralytics 8.1.35 simplify network modules by @Laughing-q in ultralytics/ultralytics#9321
• ultralytics 8.1.36 improve train stop robustness epoch + 1 >= self.epochs by @glenn-jocher in ultralytics/ultralytics#9384
• ultralytics 8.1.37 fix empty sys.argv bug by @glenn-jocher in ultralytics/ultralytics#9390
• Improved float workspace arg for TRT exports by @zldrobit in ultralytics/ultralytics#9407
• onnxruntime cpp yolo-cls fp16 fix by @Myyura in ultralytics/ultralytics#9412
• Fix None prevKeyPoints for multi-object tracking by @mlurbur in ultralytics/ultralytics#9386
• Update parameter explanations by @fcakyon in ultralytics/ultralytics#9428
• Fix TFLite Segment mask rows 1-4 normalization by @charlesollion in ultralytics/ultralytics#9405
• ultralytics 8.1.38 fix deprecated Ray Tune .is_session_enabled() by @glenn-jocher in ultralytics/ultralytics#9432
• Faster best.pt checkpoint saving by @glenn-jocher in ultralytics/ultralytics#9436
• Immediate checkpoint serialization by @glenn-jocher in ultralytics/ultralytics#9437
• Favor EMA over model in train checkpoints by @glenn-jocher in ultralytics/ultralytics#9433
• Define fallback model.names from Predictor by @glenn-jocher in ultralytics/ultralytics#9439
• Save optimizer as FP16 for smaller checkpoints by @glenn-jocher in ultralytics/ultralytics#9435
• New 🌟 per-class object counting feature and updates by @RizwanMunawar in ultralytics/ultralytics#9443
• ultralytics 8.1.39 add YOLO-World training by @Laughing-q in ultralytics/ultralytics#9268
• Fix YAML model size parsing bug by @glenn-jocher in ultralytics/ultralytics#9447
• Deprecations fix TORCH_NCCL_BLOCKING_WAIT by @glenn-jocher in ultralytics/ultralytics#9448
• Scope ultralytics/CLIP imports by @glenn-jocher in ultralytics/ultralytics#9449
• FROM pytorch/pytorch:2.2.2-cuda12.1-cudnn8-runtime by @glenn-jocher in ultralytics/ultralytics#9452
• ultralytics 8.1.40 search in Python sets {} for speed by @glenn-jocher in ultralytics/ultralytics#9450
• Fix LVIS dataset links by @glenn-jocher in ultralytics/ultralytics#9465
• Retry check_requirements once on failure by @glenn-jocher in ultralytics/ultralytics#9466
• Define export formats model.names attribute by @glenn-jocher in ultralytics/ultralytics#9474
• ultralytics 8.1.41 DDP resume untrained-checkpoint fix by @glenn-jocher in ultralytics/ultralytics#9453
• Fix TensorRT inference issue on NVIDIA Jetson by @lakshanthad in ultralytics/ultralytics#9482
• Add NVIDIA Jetson Quick Start Guide by @lakshanthad in ultralytics/ultralytics#9484
• New 🌟 Queue-Management code & docs by @RizwanMunawar in ultralytics/ultralytics#9494
• Fix Windows AttributeError: 'NoneType' object has no attribute 'encoding' by @gs80140 in ultralytics/ultralytics#9496
• Fix learning rate gap on resume by @Laughing-q in ultralytics/ultralytics#9468
• ultralytics 8.1.42 add YOLOv9 Segment models by @Burhan-Q in ultralytics/ultralytics#9296
• Add https://youtu.be/eHuzCNZeu0g to docs and integration updates by @RizwanMunawar in ultralytics/ultralytics#9525
• Add TF.js Docs Integrations Page by @abirami-vina in ultralytics/ultralytics#9512
• Solutions updates by @RizwanMunawar in ultralytics/ultralytics#9548
• Add https://youtu.be/LGGxqLZtvuw to docs & bbox dimension retrieval utilities. by @RizwanMunawar in ultralytics/ultralytics#9679
• ultralytics 8.1.43 40% faster ultralytics imports by @glenn-jocher in ultralytics/ultralytics#9547
• Fix is_online() for faster offline imports by @khoalu in ultralytics/ultralytics#9544
• Update example dataset visual by @IvorZhu331 in ultralytics/ultralytics#9559
• New is_raspberrypi function by @glenn-jocher in ultralytics/ultralytics#9826
• ultralytics 8.1.44 add IS_RASPBERRYPI and constants refactor by @glenn-jocher in ultralytics/ultralytics#9827
• Fix missing Tuner ckpt_file by @glenn-jocher in ultralytics/ultralytics#9847
• PEP660 support for pip install -e . by @glenn-jocher in ultralytics/ultralytics#9851
• ultralytics 8.1.45 fix RAM cache with mosaic buffer by @glenn-jocher in ultralytics/ultralytics#9828
• Add IS_JETSON constant by @glenn-jocher in ultralytics/ultralytics#9852
• Fix proc_device_model name for NVIDIA Jetson by @lakshanthad in ultralytics/ultralytics#9876
• Improve PNNX download robustness by @glenn-jocher in ultralytics/ultralytics#9884
• Update cla.yml by @glenn-jocher in ultralytics/ultralytics#9905
• OBB: Fix distorted plotting by @Laughing-q in ultralytics/ultralytics#9899
• Pin PNNX to avoid PNNX==20240410 bug by @glenn-jocher in ultralytics/ultralytics#9953
• ultralytics 8.1.46 add TensorRT 10 support by @Burhan-Q in ultralytics/ultralytics#9516
• Update ci.yaml for Classify-first benchmarking by @glenn-jocher in ultralytics/ultralytics#9949
• Classify datasets docs example structure update by @stormsson in ultralytics/ultralytics#9926
• Fix h5py wheel builds on arm64 Dockerfiles by @glenn-jocher in ultralytics/ultralytics#9956
• Add https://youtu.be/H3qL8ImCSV8 to docs and dataset updates by @RizwanMunawar in ultralytics/ultralytics#9990
• ultralytics 8.1.47 PNNX release 20240226 by @glenn-jocher in ultralytics/ultralytics#9954
• Add https://youtu.be/SArFQs6CHwk to docs by @RizwanMunawar in ultralytics/ultralytics#10014
• Fix increment +=2 bug in object_counter.py by @mianusman618 in ultralytics/ultralytics#9998
• Create merge-main-into-prs.yml by @glenn-jocher in ultralytics/ultralytics#10031
• Update merge-main-into-prs.yml by @glenn-jocher in ultralytics/ultralytics#10032
• Fix overrides training cfg bug by @DseidLi in ultralytics/ultralytics#10002
• Update hub.ipynb by @glenn-jocher in ultralytics/ultralytics#10062
• Update HUB notebook URLs by @glenn-jocher in ultralytics/ultralytics#10063
• HUB checkpoint upload fix by @glenn-jocher in ultralytics/ultralytics#10088
• Update Docs for YOLO-World Training ✅ by @glenn-jocher in ultralytics/ultralytics#10058
• Parameterize kpts plotting conf=0.25 by @datstat-consulting in ultralytics/ultralytics#10044
• Improve PNNX assets check robustness by @glenn-jocher in ultralytics/ultralytics#10089
• Check Google DNS for is_online() by @glenn-jocher in ultralytics/ultralytics#10095
• Raspberry Pi 5 self-hosted CI by @lakshanthad in ultralytics/ultralytics#8828

New Contributors

• @gvzdv made their first contribution in ultralytics/ultralytics#7598
• @Alarmod made their first contribution in ultralytics/ultralytics#7821
• @DieterTimmers made their first contribution in ultralytics/ultralytics#7952
• @Youho99 made their first contribution in ultralytics/ultralytics#8016
• @0xSynapse made their first contribution in ultralytics/ultralytics#8120
• @lakshanthad made their first contribution in ultralytics/ultralytics#8311
• @alexsu52 made their first contribution in ultralytics/ultralytics#8318
• @hi-sushanta made their first contribution in ultralytics/ultralytics#8285
• @AlainSchoebi made their first contribution in ultralytics/ultralytics#8368
• @kaikai23 made their first contribution in ultralytics/ultralytics#8447
• @birdcolour made their first contribution in ultralytics/ultralytics#8525
• @omarzd made their first contribution in ultralytics/ultralytics#8502
• @ALMerrill made their first contribution in ultralytics/ultralytics#8429
• @hdnh2006 made their first contribution in ultralytics/ultralytics#8648
• @bobyard-com made their first contribution in ultralytics/ultralytics#8532
• @Y-T-G made their first contribution in ultralytics/ultralytics#8629
• @massimiliano96 made their first contribution in ultralytics/ultralytics#8808
• @DseidLi made their first contribution in ultralytics/ultralytics#9196
• @zhaoruibing made their first contribution in ultralytics/ultralytics#9245
• @higor-melo made their first contribution in ultralytics/ultralytics#9330
• @ericphanson made their first contribution in ultralytics/ultralytics#9306
• @sifan-intel made their first contribution in ultralytics/ultralytics#9309
• @zldrobit made their first contribution in ultralytics/ultralytics#9407
• @mlurbur made their first contribution in ultralytics/ultralytics#9386
• @charlesollion made their first contribution in ultralytics/ultralytics#9405
• @gs80140 made their first contribution in ultralytics/ultralytics#9496
• @khoalu made their first contribution in ultralytics/ultralytics#9544
• @IvorZhu331 made their first contribution in ultralytics/ultralytics#9559
• @stormsson made their first contribution in ultralytics/ultralytics#9926
• @mianusman618 made their first contribution in ultralytics/ultralytics#9998
• @datstat-consulting made their first contribution in ultralytics/ultralytics#10044

Full Changelog: ultralytics/ultralytics@v8.1.0...v8.2.0