training_extensions
Build, train, optimize, and run computer vision models locally, from raw images to live inference. Open source, optimized for Intel XPU (CPU-only and CUDA also supported).
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 11 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/open-edge-platform/geti with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/open-edge-platform/geti'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/openvinotoolkit--training-extensions)HTML
<a href="/listings/openvinotoolkit--training-extensions"><img src="/api/listings/openvinotoolkit--training-extensions/badge.svg" alt="findsafeskills safety rating" /></a>Install
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
README.md · README · 361 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- <!-- markdownlint-disable MD013 MD033 MD041 MD042 -->
- <div align="center">
- <img src="assets/geti-header.png" alt="Geti™ - A framework to rapidly build and deploy computer vision AI models">
- <br>
- [](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml)
- [](https://codecov.io/gh/open-edge-platform/geti)
- [](https://securityscorecards.dev/viewer/?uri=github.com/open-edge-platform/geti)
- [](https://opensource.org/licenses/Apache-2.0)
- [](https://pypi.org/project/getitune/)
- [](https://clickpy.clickhouse.com/dashboard/getitune)
- **Application**: [Install](application/docs/install.md) • [Upgrade](application/docs/upgrade.md) • [Docs](https://docs.geti.intel.com/) • [License](#license)
- <br>
- **Library** (`getitune`): [Install](https://docs.geti.intel.com/docs/user-guide/library/get-started/installation) • [Docs](https://docs.geti.intel.com/docs/user-guide/library/get-started/intro)
- </div>
- **_Geti™_** is an open-source, end-to-end platform for building and continuously improving your own computer vision models. It covers annotation, fine-tuning, optimization, deployment, and inference — with 1,600+ model variants for detection, classification, and segmentation, optimized by OpenVINO™ for Intel® CPUs, GPUs, and NPUs.
- <p align="center">
- <img src="assets/model-lifecycle-infinity-light.png" width="600" alt="Geti™ - Learning Cycle"/>
- </p>
- The Geti™ application is powered by **_getitune_**, its engine for model training and optimization, which is also available standalone as a Python library. Choose the application for a no-code experience, or the `getitune` Python API for code-first workflows — both paths lead to production-ready models. Both are developed in this repository, under [`application/`](application/) and [`libraries/getitune/`](libraries/getitune/).
- ## Full lifecycle with an intuitive SME interface
- The Geti™ application runs as a native Windows app or a pre-built Docker image for Linux.
- - Install as a [**Windows application (MSIX)**](application/docs/install.md#windows-app)
- - Run as a [**Docker container**](application/docs/install.md#run-with-docker)
- - Install and launch via a [**script**](application/docs/install.md#install-script)
- - Build and run [**from source**](application/docs/install.md#run-from-source-for-development) (for development)
- For complete, step-by-step instructions - including prerequisites, GPU/accelerator support, container and source builds, the install script, and troubleshooting - see the [Installation guide](application/docs/install.md). To update an existing Geti™ installation to a newer version, follow the [Upgrade guide](application/docs/upgrade.md).
- <p align="center">
- <img src="assets/geti-application-demo.gif" alt="Build and optimize a model in the Geti™ application" width="80%">
- </p>
- > [!TIP]
- > The [documentation](https://docs.geti.intel.com/) is a valuable resource to learn more about Geti™ and its capabilities.
- > New users are encouraged to read the step-by-step tutorial ["Train your first model"](https://docs.geti.intel.com/docs/user-guide/quick-start/training-your-first-model).
- ## Full lifecycle with a simple Python API
- Geti's training engine is also available on PyPI as `getitune`, an open-source Python library for training, optimizing, and deploying models.
- <p align="center">
- <img src="libraries/getitune/assets/getitune-demo.gif" alt="Train and optimize a model with the getitune Python API" width="80%">
- </p>
- <details>
- <summary>Install <code>getitune</code> and run the demo code</summary>
- ```bash
- uv pip install "getitune[xpu]" --extra-index-url https://download.pytorch.org/whl/xpu # for Intel® XPU acceleration
- uv pip install "getitune[cuda]" --extra-index-url https://download.pytorch.org/whl/cu128 # for NVIDIA® CUDA acceleration
- uv pip install getitune # CPU-only by default
- ```
- > [!IMPORTANT]
- > **Ultralytics YOLO models** are NOT included in the package by default. To install these models, which are distributed
- > under the [AGPL-3.0 license](./THIRD-PARTY-NOTICES.md#ultralytics), you must specify the `ultralytics` extra as
- > explained in the [getitune documentation](https://docs.geti.intel.com/docs/user-guide/library/get-started/installation).
- **Pick a model, then train, export, and quantize it:**
- ```python
- from getitune.engine import create_engine
- from getitune.utils import list_models
- # Explore available models to pick the one you need
- all_models = list_models() # All model names
- detection_models = list_models(task="DETECTION") # Filter by task
- recipes = list_models(return_recipes=True) # Full recipe YAML paths
- # Choose a model and point it at your dataset
- engine = create_engine(
- model="edgecrafter_m",
- data="./led_counting",
- work_dir="./my_workspace", # checkpoints, exports, and logs
- device="auto", # "auto", "cpu", "gpu", or "xpu"
- )
- # Train the model on your own Intel CPU or GPU
- engine.train()
- # Export the trained model to OpenVINO format (.xml + .bin)
- exported_model_path = engine.export()
- # Load the exported model into the OpenVINO engine
- openvino_engine = create_engine(model=exported_model_path, data=engine.datamodule)
- # Quantize the model to INT8, optimized for Intel hardware
- optimized_model_path = openvino_engine.optimize()
- # Run inference with the optimized INT8 model
- predictions = openvino_engine.predict(checkpoint=optimized_model_path)
- ```
- See the [getitune documentation](https://docs.geti.intel.com/docs/user-guide/library/get-started/quick-start) for the full list of recipes,
- advanced configuration, dataset support, backend-specific options, and deployment/optimization examples.
- </details>
- ## Key Features
- <details open>
- <summary>🏆 State-of-the-art model catalog</summary>
- Train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, YOLOX, D-FINE, and Mask R-CNN.
- Would you like to see a specific model added? Let us know by opening a [GitHub issue](https://github.com/open-edge-platform/geti/issues)!
- <!-- markdownlint-disable MD060 -->
- <table>
- <thead>
- <tr>
- <th width="30%">Computer Vision Task</th>
- <th>Model Architecture</th>
- <th>Paper</th>
- </tr>
- </thead>
- <tbody>
- <tr>
- <td rowspan="12"><b>Object Detection</b><br>Locate and classify objects with bounding boxes. Common use cases: counting items, defect localization, surveillance.</td>
- <td>D-FINE M / L / X</td>
- <td><a href="https://arxiv.org/abs/2412.04234">DEIM</a> + <a href="https://arxiv.org/abs/2410.13842">D-FINE</a></td>
- </tr>
- <tr>
- <td>DINOv3 DETR S / M / L</td>
- <td><a href="https://arxiv.org/abs/2508.10104">DINOv3</a> + <a href="https://arxiv.org/html/2509.20787v4">DEIMv2</a> + <a href="https://arxiv.org/abs/2005.12872">DETR</a></td>
- </tr>
- <tr>
- <td>ECDet S / M / L / X</td>
- <td><a href="https://arxiv.org/abs/2603.18739">EdgeCrafter</a></td>
- </tr>
- <tr>
- <td>MobileNet V2 ATSS</td>
- <td><a href="https://arxiv.org/abs/1801.04381">MobileNetV2</a> + <a href="https://arxiv.org/abs/1912.02424">ATSS</a></td>
- </tr>
- <tr>
- <td>MobileNet V2 SSD</td>
- <td><a href="https://arxiv.org/abs/1801.04381">MobileNetV2</a> + <a href="https://arxiv.org/abs/1512.02325">SSD</a></td>
- </tr>
- <tr>
- <td>RF-DETR N / S / M / L</td>
- <td><a href="https://arxiv.org/abs/2511.09554">RF-DETR</a></td>
- </tr>
- <tr>
- <td>RT-DETR R50</td>
- <td><a href="https://arxiv.org/abs/2304.08069">RT-DETR</a></td>
- </tr>
- <tr>
- <td>RT-DETRv2 R18 / R34 / R50 / R101</td>
- <td><a href="https://arxiv.org/abs/2407.17140">RT-DETRv2</a></td>
- </tr>
- <tr>
- <td>YOLO11 N / S / M / L / X</td>
- <td><a href="https://docs.ultralytics.com/models/yolo11#overview">Ultralytics YOLO11</a></td>
- </tr>
- <tr>
- <td>YOLO12 N / S / M / L / X</td>
- <td><a href="https://arxiv.org/abs/2502.12524">YOLOv12</a></td>
- </tr>
- <tr>
- <td>YOLO26 N / S / M / L / X</td>
- <td><a href="https://arxiv.org/abs/2606.03748">YOLO26</a></td>
- </tr>
- <tr>
- <td>YOLOX Tiny / S / L / X</td>
- <td><a href="https://arxiv.org/abs/2107.08430">YOLOX</a></td>
- </tr>
- <tr>
- <td rowspan="10"><b>Instance Segmentation</b><br>Detect objects and produce pixel-precise masks per instance. Common use cases: medical imaging, robotics, area estimation.</td>
- <td>EoMT Large 640</td>
- <td><a href="https://arxiv.org/abs/2503.19108">EoMT</a></td>
- </tr>
- <tr>
- <td>EoMT-DINOv3 Large 640</td>
- <td><a href="https://arxiv.org/abs/2508.10104">DINOv3</a> + <a href="https://arxiv.org/abs/2503.19108">EoMT</a></td>
- </tr>
- <tr>
- <td>Mask2Former Swin-T / S / B / L</td>
- <td><a href="https://arxiv.org/abs/2112.01527">Mask2Former</a></td>
- </tr>
- <tr>
- <td>Mask-RCNN EfficientNet B2</td>
- <td><a href="https://arxiv.org/abs/1905.11946">EfficientNet</a> + <a href="https://arxiv.org/abs/1703.06870">Mask R-CNN</a></td>
- </tr>
- <tr>
- <td>Mask-RCNN ResNet50</td>
- <td><a href="https://arxiv.org/abs/1512.03385">ResNet</a> + <a href="https://arxiv.org/abs/1703.06870">Mask R-CNN</a></td>
- </tr>
- <tr>
- <td>Mask-RCNN Swin-T</td>
- <td><a href="https://arxiv.org/abs/2103.14030">Swin Transformer</a> + <a href="https://arxiv.org/abs/1703.06870">Mask R-CNN</a></td>
- </tr>
- <tr>
- <td>RF-DETR N / S / M / L / XL / 2XL</td>
- <td><a href="https://arxiv.org/abs/2511.09554">RF-DETR</a></td>
- </tr>
- <tr>
- <td>RTMDet Tiny</td>
- <td><a href="https://arxiv.org/abs/2212.07784">RTMDet</a></td>
- </tr>
- <tr>
- <td>YOLO11 N / S / M / L / X</td>
- <td><a href="https://docs.ultralytics.com/models/yolo11#overview">Ultralytics YOLO11</a></td>
- </tr>
- <tr>
- <td>YOLO26 N / S / M / L / X</td>
- <td><a href="https://arxiv.org/abs/2606.03748">YOLO26</a></td>
- </tr>
- <tr>
- <td rowspan="10"><b>Classification</b> (multi-class, multi-label)<br>Assign one or more labels to an entire image. Common use cases: defect classification, product categorization, content tagging.</td>
- <td>ConvNeXt T / S / B / L</td>
- <td><a href="https://arxiv.org/abs/2201.03545">ConvNeXt</a></td>
- </tr>
- <tr>
- <td>ConvNeXt V2 A / B / L</td>
- <td><a href="https://arxiv.org/abs/2301.00808">ConvNeXt V2</a></td>
- </tr>
- <tr>
- <td>DINOv3 ViT S / B / L</td>
- <td><a href="https://arxiv.org/abs/2508.10104">DINOv3</a></td>
- </tr>
- <tr>
- <td>DINOv2 Small</td>
- <td><a href="https://arxiv.org/abs/2304.07193">DINOv2</a></td>
- </tr>
- <tr>
- <td>EfficientNet B0 / B3</td>
- <td><a href="https://arxiv.org/abs/1905.11946">EfficientNet</a></td>
- </tr>
- <tr>
- <td>EfficientNet V2 Small</td>
- <td><a href="https://arxiv.org/abs/2104.00298">EfficientNetV2</a></td>
- </tr>
- <tr>
- <td>MobileNet V3 Large</td>
- <td><a href="https://arxiv.org/abs/1905.02244">MobileNetV3</a></td>
- </tr>
- <tr>
- <td>ViT Tiny</td>
- <td><a href="https://arxiv.org/abs/2010.11929">ViT</a></td>
- </tr>
- <tr>
- <td>YOLO26 N / S / M / L / X</td>
- <td><a href="https://arxiv.org/abs/2606.03748">YOLO26</a></td>
- </tr>
- <tr>
- <td>Other models from <i>timm</i> (1600+ backbones)</td>
- <td><a href="https://github.com/huggingface/pytorch-image-models">timm</a></td>
- </tr>
- </tbody>
- </table>
- <!-- markdownlint-enable MD060 -->
- </details>
- <details>
- <summary>🔄 Interactive end-to-end model training</summary>
- Geti™ enables users to start building deep-learning computer vision models with as few as 10-20 images and take them to production in one environment - annotate, train, optimize, run inference, and improve accuracy in a rapid train-predict-annotate loop.
- </details>
- <details>
- <summary>⚡ Hardware-accelerated inference & model optimization</summary>
- Every model is automatically exported with [OpenVINO™](https://www.intel.com/content/www/us/en/developer/tools/openvino-toolkit/overview.html) for deployment across the full Intel® XPU portfolio (Arc™ GPUs, Core™ Ultra processors); NVIDIA® CUDA and CPU-only execution are also supported. Fine-tune and run inference directly on edge and client hardware - including Intel® Panther Lake and Arc™ Battlemage (B-series) GPUs - with no Kubernetes cluster or data-center GPU required. Built-in accuracy-aware INT8 quantization further reduces model size and latency on resource-constrained edge devices with minimal impact on accuracy.
- </details>
- <details>
- <summary>🚀 Integrated deployment & inference</summary>
- Build custom pipelines (source → model → sink) to deploy models inside Geti and monitor real-time predictions on video streams. Sources include USB/IP cameras and video files; optional sinks include folder, MQTT, and webhook. Complete pipelines can be exported as OpenVINO™-optimized bundles for edge deployment.
- </details>
- <details>
- <summary>🎨 Multiple computer vision tasks</summary>
- Geti™ supports [multiple computer vision tasks](https://docs.geti.intel.com/docs/user-guide/learn-geti/computer-vision-tasks/ai-fundamentals-tasks) that are commonly employed across various use cases - image classification, object detection and instance segmentation from the no-code web interface, with even more tasks available through the `getitune` library.
- </details>
- <details>
- <summary>🧠 Smart annotations</summary>
- Smart annotations in Geti™ enable users to easily create bounding boxes and polygons. These smart annotation features coupled with the AI-assisted annotations and state-of-the-art AI models such as the Segment Anything Model keep human experts in the loop while massively reducing the total annotation efforts needed by a human.
- <p align="center">
- <img src="application/docs/assets/annotator.webp" width="600" alt="Smart Annotations">
- </p>
- </details>
- <details>
- <summary>📦 Model & dataset management</summary>
- Track how datasets and models evolve, link models to a specific dataset revision, view exact training hyperparameters, and fine-tune from any previous version. Import and export in COCO, Pascal VOC, YOLO, and a Geti-optimized native format, with label filtering to selectively include or exclude labels on import/export.
- </details>
- ## Ecosystem
- Explore other open-source AI projects by Intel®:
- - [Anomalib](https://github.com/open-edge-platform/anomalib) - An anomaly detection suite comprising state-of-the-art algorithms and features such as experiment management, hyper-parameter optimization and edge inference.
- - [Instant Learn](https://github.com/open-edge-platform/instant-learn) - A framework for developing, benchmarking, and deploying zero-shot visual prompting algorithms on the edge.
- - [Datumaro](https://github.com/open-edge-platform/datumaro) - Dataset Management Framework, a Python library and a CLI tool to build, analyze and manage Computer Vision datasets.
- - [OpenVINO™](https://github.com/openvinotoolkit/openvino) - Software toolkit for optimizing and deploying deep learning models.
- - [OpenVINO™ Model Server](https://github.com/openvinotoolkit/model_server) - A scalable inference server for models optimized with OpenVINO™.
- - [Model API](https://github.com/open-edge-platform/model_api) - A set of wrapper classes for particular tasks and model architectures, simplifying data preprocessing and postprocessing as well as routine procedures.
- - [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) - An end-to-end framework for teaching robots to perform tasks through imitation learning from human demonstrations.
- ## Who uses Geti™?
- Geti™ is used by research institutes, industrial partners, universities and AI enthusiasts. Applications range from robotics to medical analysis or industrial quality control. There are some interesting members of the Geti™ community:
- - [Intel Foundry](https://medium.com/open-edge-platform/solving-silicon-foundry-woes-with-ai-vision-geti-and-a-robotic-dog-a8382b5d9267)
- - [Royal Brompton and Harefield hospitals](https://www.rbht.nhs.uk/artificial-intelligence-theme-new-trust-led-research)
- - [WSC Sports](https://www.linkedin.com/posts/wsc-sports-technologies_revolutionizing-sports-broadcasting-with-activity-7161419649878773761-cUM3/)
- - [Dell NativeEdge](https://infohub.delltechnologies.com/en-us/p/transforming-edge-ai-with-continuous-learning-meet-intel-geti-and-openvino-on-dell-nativeedge/)
- - [Bravent](https://www.linkedin.com/posts/bravent_intelgeti-openvino-manufacturing-activity-7214544905086390272-H19g/)
- - [ASRock Industrial](https://www.asrockind.com/en-gb/article/176)
- - [PeopleSense.AI](https://community.intel.com/t5/Blogs/Tech-Innovation/Artificial-Intelligence-AI/Intel-Liftoff-Days-2024-Highlights-from-the-Third-Edition/post/1661265)
- - [Capgemini](https://www.capgemini.com/insights/expert-perspectives/capgemini-and-intel-corporation-redefining-the-future-of-robotics-and-physical-ai/)
- ## Contribute
- To report a bug or submit a feature request, please open a [GitHub issue](https://github.com/open-edge-platform/geti/issues). If you have an open question, ask in [GitHub Discussions](https://github.com/open-edge-platform/geti/discussions).
- For developers who would like to contribute with a pull request, see the [Contributing guide](CONTRIBUTING.md) for details.
- <p align="center">
- <b>Thank you 👏 to all our contributors!</b>
- </p>
- <a href="https://github.com/open-edge-platform/geti/graphs/contributors">
- <img src="https://contrib.rocks/image?repo=open-edge-platform/geti" alt="Contributors" />
- </a>
- ## License
- Geti™ source code is licensed under the [Apache License Version 2.0](LICENSE).
- Docker images contain Geti™ components licensed under Apache 2.0 and may also contain third-party components governed by
- different license terms, listed separately in [this file](THIRD-PARTY-NOTICES.md).
- The Windows MSIX App is licensed under the [Intel Simplified Software License](https://software.intel.com/sites/landingpage/pintool/intel-simplified-software-license.txt).
- Regarding models trained with Geti™, their license depends on the model type and is indicated within the application;
- most are licensed under Apache 2.0, while some fall under more restrictive licenses (e.g. [Ultralytics AGPL3.0 models](./THIRD-PARTY-NOTICES.md#ultralytics)).
- It is the end user's responsibility to ensure that the specific use of Geti™ and the models complies with the respective licenses.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.