From 3892015406e65b4735f82957fc87185482bdbb01 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 14:50:01 +0100 Subject: [PATCH 01/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 256 +++++++++++++++++++++++++++--------------------------- 1 file changed, 130 insertions(+), 126 deletions(-) diff --git a/README.md b/README.md index 6a9a2c901b..ed36551d2c 100644 --- a/README.md +++ b/README.md @@ -3,28 +3,22 @@ Geti™ - A framework to rapidly build and deploy computer vision AI models -**Enable anyone from domain experts to data scientists to rapidly develop production-ready AI models** - -[Key Features](#key-features) • -[Supported tasks and models](#supported-tasks-and-models) • [Quick Start](#quick-start) • -[Documentation](#documentation) • -[Community](#community) +[Geti™ documentation](https://docs.geti.intel.com/) • +[`getitune` documentation](https://open-edge-platform.github.io/geti/latest/index.html) -[![Daily checks](https://github.com/open-edge-platform/training_extensions/actions/workflows/daily.yml/badge.svg)](https://github.com/open-edge-platform/training_extensions/actions/workflows/daily.yml) -[![Docker build](https://github.com/open-edge-platform/training_extensions/actions/workflows/build.yaml/badge.svg)](https://github.com/open-edge-platform/training_extensions/actions/workflows/build.yaml) -[![Codecov](https://codecov.io/gh/open-edge-platform/training_extensions/branch/develop/graph/badge.svg?token=9HVFNMPFGD)](https://codecov.io/gh/open-edge-platform/training_extensions) -[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/open-edge-platform/training_extensions/badge)](https://securityscorecards.dev/viewer/?uri=github.com/open-edge-platform/training_extensions) +[![Container build](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml/badge.svg)](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml) +[![Daily checks](https://github.com/open-edge-platform/geti/actions/workflows/daily.yml/badge.svg)](https://github.com/open-edge-platform/geti/actions/workflows/daily.yml) +[![Codecov](https://codecov.io/gh/open-edge-platform/geti/branch/develop/graph/badge.svg?token=9HVFNMPFGD)](https://codecov.io/gh/open-edge-platform/geti) +[![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/open-edge-platform/geti/badge)](https://securityscorecards.dev/viewer/?uri=github.com/open-edge-platform/geti) [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0) +[![PyPI version](https://img.shields.io/pypi/v/getitune?logo=pypi&logoColor=white)](https://pypi.org/project/getitune/) +[![PyPI downloads](https://static.pepy.tech/badge/getitune)](https://clickpy.clickhouse.com/dashboard/getitune) -## Introduction - -Geti™ is an end-to-end application that takes you from raw images to a deployed computer vision model — annotate, train, -optimize, and run inference, all in one place, all on your own hardware. Start with as few as 10-20 images and iterate -in a rapid, feedback-driven loop. Geti runs locally as a single Docker image or a native Windows app, and is optimized -for Intel® hardware with OpenVINO™ for fast inference across the full Intel® XPU portfolio. +Geti™ is an end-to-end Vision AI application that takes you from raw images to a deployed computer vision model - annotate, train, +optimize and run inference - all in one place, all on your own hardware. Geti™ runs locally as a single container or a native Windows app. Geti™ is optimized for Intel® hardware with OpenVINO™ for fast inference across the full Intel® XPU portfolio.

Geti™ - Learning Cycle @@ -39,6 +33,99 @@ for Intel® hardware with OpenVINO™ for fast inference across the full Intel® > > The development of the Geti™ application now continues in this repository in the [`application`](application) folder. > Previous versions of Geti™ are still available in a separate [repository](https://github.com/open-edge-platform/geti_v2). +> In general, we recommend upgrading to the latest Geti™ release whenever possible - not only to access new functionality, +> but also to receive better support from Intel and the Geti™ community. +> For upgrade from Geti™ v2 to v3, please follow the [upgrade guidance](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/migration-from-geti-2x). + + +# Quick Start + +Get Geti running and train your first model in a few minutes. For full instructions and all options, see the +[official documentation](https://docs.geti.intel.com/) and the [application README](application/README.md). + +**Minimum recommended setup:** 8 CPU threads, 16 GB RAM, 40 GB free disk. A GPU (Intel® XPU or NVIDIA® CUDA) is +recommended for larger models. + +## 1. Run Geti + +### Windows Application + +Run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments. + +Download the Windows Installer: + +- [Download CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) +- [Download Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) +- [Download Nvidia® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) + +Install Geti Windows application and launch it from the Start menu + +### Docker + +Pull a pre-built image for your hardware and launch it: + +```bash +docker pull ghcr.io/open-edge-platform/geti-xpu # modern Intel® CPU/GPU (recommended) +docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® CUDA platforms +docker pull ghcr.io/open-edge-platform/geti-cpu # CPU-only (most lightweight) + +# Retag the pulled image as `geti-{cpu,xpu,cuda}:latest` for using with `just run-image` +docker tag ghcr.io/open-edge-platform/geti-cpu:latest geti-cpu:latest + +just run-image --accelerator xpu # launch the application +``` + +Then open the Geti web application at [**http://localhost:7860**](http://localhost:7860). + +For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the +[application README](application/README.md). + +#### Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export + +Linux, WSL (In order to run a script you need to have curl & git installed): + +```bash +curl -fsSL https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.sh | bash +``` + +## 2. Train your first model + +Once Geti is running, build your first model directly in the web UI: + +1. **Create a project** — choose a task (object detection, instance segmentation, or classification) and define your labels. +2. **Upload media** — drag in 20–50 representative images to start. +3. **Annotate** — label your media with the built-in manual and AI-assisted tools. +4. **Train** — pick a recommended architecture and start training; watch progress in the Jobs panel. +5. **Deploy** — build an inference pipeline (source → model → sink) and run predictions in real time, or export an + OpenVINO™-optimized bundle for the edge. + +See [Training your first model](https://docs.geti.intel.com/) for the full walkthrough. + +### Use the Python API (`getitune`) + +Prefer to work programmatically? Geti's training engine is published on PyPI and can train, optimize, and deploy models +from Python. It requires **Python 3.11–3.14**, **PyTorch 2.10**, **OpenVINO™ 2026.1**, and **NumPy ≥ 2.0**. + +```bash +pip install "getitune[cpu]" # or [xpu] for Intel® GPU, [cuda] for NVIDIA® GPU +``` + +```python +from getitune.engine import create_engine + +# Initialize and train using a bundled recipe and dataset +engine = create_engine( + data="tests/assets/classification_cifar10", + model="src/getitune/recipe/classification/multi_class_cls/efficientnet_b0.yaml", +) +engine.train() +engine.test() +exported_path = engine.export() # writes OpenVINO IR +``` + +See the [library README](library/README.md) for the full list of recipes, advanced configuration, dataset support, and +inference/optimization examples. + ## Key Features @@ -72,7 +159,7 @@ for Intel® hardware with OpenVINO™ for fast inference across the full Intel® Below is a list of tasks and model architectures supported by Geti™. Some tasks are available directly from the no-code web interface, while others are accessible through the Python API (`getitune`) — both are part of the same Geti application. -Would you like to see a specific model added to the list? Let us know by opening a [GitHub issue](https://github.com/open-edge-platform/training_extensions/issues)! +Would you like to see a specific model added to the list? Let us know by opening a [GitHub issue](https://github.com/open-edge-platform/geti/issues)! @@ -167,127 +254,44 @@ Would you like to see a specific model added to the list? Let us know by opening -> [!TIP] -> Other projects of the Open Edge Platform enable even more tasks and models, check them: -> -> - [Anomalib (Studio)](https://github.com/open-edge-platform/anomalib) → anomaly detection -> - [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) → robot learning, VLA (Vision-Language-Action) -> - [Instant Learn](https://github.com/open-edge-platform/instant-learn) → visual prompting -> - [OpenVINO™](https://github.com/openvinotoolkit/openvino) - Software toolkit for optimizing and deploying deep learning models. -> - [Model API](https://github.com/open-edge-platform/model_api) - wrapper that simplifies model loading, execution, and data processing for easy inference - -## Quick Start - -Get Geti running and train your first model in a few minutes. For full instructions and all options, see the -[official documentation](https://docs.geti.intel.com/) and the [application README](application/README.md). - -**Minimum recommended setup:** 8 CPU threads, 16 GB RAM, 40 GB free disk. A GPU (Intel® XPU or NVIDIA® CUDA) is -recommended for larger models. - -### 1. Run Geti - -#### Windows Application - -Run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments. -Download the Windows Installer: +## Ecosystem -- [Download CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) -- [Download Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) -- [Download Nvidia® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) +- [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) → visual prompting +- [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) → robot learning, VLA (Vision-Language-Action) +- [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 preprocess and postprocess as well as routine procedures. -Install Geti Windows application and launch it from the Start menu +## Who uses Geti™? -#### Docker +Geti™ is a powerful tool to build vision models for a wide range of processes, including detecting defective parts in a production line, reducing downtime on the factory floor, automating inventory management, or other automation projects. We have chosen to highlight a few interesting community members: -Pull a pre-built image for your hardware and launch it: +- [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/) -```bash -docker pull ghcr.io/open-edge-platform/geti-xpu # modern Intel® CPU/GPU (recommended) -docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® CUDA platforms -docker pull ghcr.io/open-edge-platform/geti-cpu # CPU-only (most lightweight) - -# Retag the pulled image as `geti-{cpu,xpu,cuda}:latest` for using with `just run-image` -docker tag ghcr.io/open-edge-platform/geti-cpu:latest geti-cpu:latest - -just run-image --accelerator xpu # launch the application -``` - -Then open the Geti web application at [**http://localhost:7860**](http://localhost:7860). - -For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the -[application README](application/README.md). - -#### Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export - -Linux, WSL (In order to run a script you need to have curl & git installed): - -```bash -curl -fsSL https://raw.githubusercontent.com/open-edge-platform/training_extensions/develop/install.sh | bash -``` - -### 2. Train your first model - -Once Geti is running, build your first model directly in the web UI: - -1. **Create a project** — choose a task (object detection, instance segmentation, or classification) and define your labels. -2. **Upload media** — drag in 20–50 representative images to start. -3. **Annotate** — label your media with the built-in manual and AI-assisted tools. -4. **Train** — pick a recommended architecture and start training; watch progress in the Jobs panel. -5. **Deploy** — build an inference pipeline (source → model → sink) and run predictions in real time, or export an - OpenVINO™-optimized bundle for the edge. - -See [Training your first model](https://docs.geti.intel.com/) for the full walkthrough. - -### Use the Python API (`getitune`) - -Prefer to work programmatically? Geti's training engine is published on PyPI and can train, optimize, and deploy models -from Python. It requires **Python 3.11–3.14**, **PyTorch 2.10**, **OpenVINO™ 2026.1**, and **NumPy ≥ 2.0**. - -```bash -pip install "getitune[cpu]" # or [xpu] for Intel® GPU, [cuda] for NVIDIA® GPU -``` - -```python -from getitune.engine import create_engine - -# Initialize and train using a bundled recipe and dataset -engine = create_engine( - data="tests/assets/classification_cifar10", - model="src/getitune/recipe/classification/multi_class_cls/efficientnet_b0.yaml", -) -engine.train() -engine.test() -exported_path = engine.export() # writes OpenVINO IR -``` - -See the [library README](library/README.md) for the full list of recipes, advanced configuration, dataset support, and -inference/optimization examples. - -## Migrating from Geti 2.x - -Geti 3.0 introduces a simplified dataset‑based workflow: datasets must be exported and imported individually, models from 2.x require retraining, project-level migration is replaced by dataset-level transfer, and the REST API and deployment now use the OpenVINO™ Model API — **Please follow the -[migration guidance](https://docs.geti.intel.com/) in the documentation.** - -## Documentation - -For complete user and developer documentation, visit [**docs.geti.intel.com**](https://docs.geti.intel.com/). - -| Component | README | Documentation | -| ------------------------- | ---------------------------------------------- | ---------------------------------------------------------------------------------- | -| **Geti application** | [application/README.md](application/README.md) | [docs.geti.intel.com](https://docs.geti.intel.com/) | -| **Python API (getitune)** | [library/README.md](library/README.md) | [Docs](https://open-edge-platform.github.io/training_extensions/latest/index.html) | ## Community -- To report a bug or submit a feature request, please open a [GitHub issue](https://github.com/open-edge-platform/training_extensions/issues). -- Ask questions via [GitHub Discussions](https://github.com/open-edge-platform/training_extensions/discussions). +- To report a bug or submit a feature request, please open a [GitHub issue](https://github.com/open-edge-platform/geti/issues). +- Ask questions via [GitHub Discussions](https://github.com/open-edge-platform/geti/discussions). + +## Contribute -For those who would like to contribute, see [CONTRIBUTING.md](CONTRIBUTING.md) for details. +For those who would like to contribute, see [Contributing guide](CONTRIBUTING.md) for details. -Thank you! We appreciate your support! +

+ Thank you 👏 to all our contributors! +

- + Contributors From 4acc5da6fd8a2728431c583653bef5560757f964 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 15:33:52 +0100 Subject: [PATCH 02/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 108 ++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 72 insertions(+), 36 deletions(-) diff --git a/README.md b/README.md index ed36551d2c..b81decf8ff 100644 --- a/README.md +++ b/README.md @@ -8,7 +8,6 @@ [`getitune` documentation](https://open-edge-platform.github.io/geti/latest/index.html) [![Container build](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml/badge.svg)](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml) -[![Daily checks](https://github.com/open-edge-platform/geti/actions/workflows/daily.yml/badge.svg)](https://github.com/open-edge-platform/geti/actions/workflows/daily.yml) [![Codecov](https://codecov.io/gh/open-edge-platform/geti/branch/develop/graph/badge.svg?token=9HVFNMPFGD)](https://codecov.io/gh/open-edge-platform/geti) [![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/open-edge-platform/geti/badge)](https://securityscorecards.dev/viewer/?uri=github.com/open-edge-platform/geti) [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0) @@ -17,8 +16,7 @@ -Geti™ is an end-to-end Vision AI application that takes you from raw images to a deployed computer vision model - annotate, train, -optimize and run inference - all in one place, all on your own hardware. Geti™ runs locally as a single container or a native Windows app. Geti™ is optimized for Intel® hardware with OpenVINO™ for fast inference across the full Intel® XPU portfolio. +Geti™ is an end-to-end Vision AI application that takes you from raw images to a deployed computer vision model. Geti™ runs locally as a single container or a native Windows app and is optimized for fast inference across the full Intel® XPU portfolio.

Geti™ - Learning Cycle @@ -127,46 +125,29 @@ See the [library README](library/README.md) for the full list of recipes, advanc inference/optimization examples. + + + ## Key Features -- **Interactive end-to-end model training**: 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. -- **State-of-the-art model catalog**: train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, - YOLOX, D-FINE, and Mask R-CNN — see the [full list below](#supported-tasks-and-models). -- **Multiple computer vision tasks**: image classification, object detection, and instance segmentation from the no-code - web interface, with even more tasks available through the Python API (`getitune`). -- **Smart annotations**: manual and semi-automated labeling powered by models like SAM (Segment Anything Model), plus - bulk labeling to dramatically speed up dataset creation. -- **Dataset & model versioning**: track how datasets and models evolve, link models to a specific dataset revision, view - exact training hyperparameters, and fine-tune from any previous version. -- **Runs locally, on the edge**: fine-tune models 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. - Minimum recommended setup: **8 CPU threads, 16 GB RAM, 40 GB free disk**. -- **Hardware acceleration**: optimized for modern Intel® hardware (Arc™ GPUs, Core™ Ultra processors). 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; NVIDIA® CUDA and CPU-only execution are also supported. -- **Dataset import & export**: COCO, Pascal VOC, and YOLO formats plus a Geti-optimized native format, with label - filtering to selectively include or exclude labels on import/export. -- **Model optimization**: built-in quantization with accuracy-aware INT8 optimization to balance inference speed and - accuracy on resource-constrained edge devices. -- **Integrated deployment & inference**: 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. - -## Supported tasks and models - -Below is a list of tasks and model architectures supported by Geti™. Some tasks are available directly from the no-code -web interface, while others are accessible through the Python API (`getitune`) — both are part of the same Geti -application. -Would you like to see a specific model added to the list? Let us know by opening a [GitHub issue](https://github.com/open-edge-platform/geti/issues)! +

+🔄 Interactive end-to-end model training + +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. + +
+ +
+🏆 State-of-the-art model catalog + +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)! - + @@ -252,8 +233,63 @@ Would you like to see a specific model added to the list? Let us know by opening - + + +
+🎨 Multiple computer vision tasks + +Image classification, object detection, and instance segmentation from the no-code web interface, with even more tasks available through the Python API (`getitune`). + +
+ +
+🧠 Smart annotations + +Manual and semi-automated labeling powered by models like SAM (Segment Anything Model), plus bulk labeling to dramatically speed up dataset creation. + +
+ +
+📦 Dataset & model versioning + +Track how datasets and models evolve, link models to a specific dataset revision, view exact training hyperparameters, and fine-tune from any previous version. + +
+ +
+🏔️ Runs locally, on the edge + +Fine-tune models 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. Minimum recommended setup: **8 CPU threads, 16 GB RAM, 40 GB free disk**. + +
+ +
+⚡ Hardware acceleration + +Optimized for modern Intel® hardware (Arc™ GPUs, Core™ Ultra processors). 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; NVIDIA® CUDA and CPU-only execution are also supported. + +
+ +
+💾 Dataset import & export + +COCO, Pascal VOC, and YOLO formats plus a Geti-optimized native format, with label filtering to selectively include or exclude labels on import/export. + +
+ +
+🔧 Model optimization + +Built-in quantization with accuracy-aware INT8 optimization to balance inference speed and accuracy on resource-constrained edge devices. + +
+ +
+🚀 Integrated deployment & inference + +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. +
## Ecosystem From dddf9a884e083c86e9fc043a8e069ca74f716d4f Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 15:35:32 +0100 Subject: [PATCH 03/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index b81decf8ff..b8d7d5fea0 100644 --- a/README.md +++ b/README.md @@ -147,7 +147,7 @@ Train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, Y
Computer Vision TaskComputer Vision Task Model Architecture Paper
- + From 575302f39505112f9618592d3133e55c0623e838 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 15:36:25 +0100 Subject: [PATCH 04/17] width Signed-off-by: Barabanov, Alexander --- README.md | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/README.md b/README.md index b8d7d5fea0..e29ec64c2e 100644 --- a/README.md +++ b/README.md @@ -125,9 +125,6 @@ See the [library README](library/README.md) for the full list of recipes, advanc inference/optimization examples. - - - ## Key Features
@@ -147,7 +144,7 @@ Train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, Y
Computer Vision TaskComputer Vision Task Model Architecture Paper
- + From 3b5715e967630266fc11eebf30ae68c831ba9d4b Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 16:00:44 +0100 Subject: [PATCH 05/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 90 +++++++++++++++++++++++++------------------------------ 1 file changed, 40 insertions(+), 50 deletions(-) diff --git a/README.md b/README.md index e29ec64c2e..bd7af9b538 100644 --- a/README.md +++ b/README.md @@ -11,8 +11,8 @@ [![Codecov](https://codecov.io/gh/open-edge-platform/geti/branch/develop/graph/badge.svg?token=9HVFNMPFGD)](https://codecov.io/gh/open-edge-platform/geti) [![OpenSSF Scorecard](https://api.securityscorecards.dev/projects/github.com/open-edge-platform/geti/badge)](https://securityscorecards.dev/viewer/?uri=github.com/open-edge-platform/geti) [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0) -[![PyPI version](https://img.shields.io/pypi/v/getitune?logo=pypi&logoColor=white)](https://pypi.org/project/getitune/) -[![PyPI downloads](https://static.pepy.tech/badge/getitune)](https://clickpy.clickhouse.com/dashboard/getitune) +[![PyPI version](https://img.shields.io/pypi/v/getitune?logo=pypi&logoColor=white)](https://pypi.org/project/getitune/) +[![PyPI downloads](https://static.pepy.tech/badge/getitune)](https://clickpy.clickhouse.com/dashboard/getitune) @@ -31,11 +31,10 @@ Geti™ is an end-to-end Vision AI application that takes you from raw images to > > The development of the Geti™ application now continues in this repository in the [`application`](application) folder. > Previous versions of Geti™ are still available in a separate [repository](https://github.com/open-edge-platform/geti_v2). -> In general, we recommend upgrading to the latest Geti™ release whenever possible - not only to access new functionality, +> In general, we recommend upgrading to the latest Geti™ release whenever possible - not only to access new functionality, > but also to receive better support from Intel and the Geti™ community. > For upgrade from Geti™ v2 to v3, please follow the [upgrade guidance](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/migration-from-geti-2x). - # Quick Start Get Geti running and train your first model in a few minutes. For full instructions and all options, see the @@ -44,6 +43,12 @@ Get Geti running and train your first model in a few minutes. For full instructi **Minimum recommended setup:** 8 CPU threads, 16 GB RAM, 40 GB free disk. A GPU (Intel® XPU or NVIDIA® CUDA) is recommended for larger models. +| Component | Minimum requirement | +| --------- | ------------------- | +| CPU | 8 threads | +| RAM | 16 GB | +| Disk | 40 GB free | + ## 1. Run Geti ### Windows Application @@ -56,7 +61,7 @@ Download the Windows Installer: - [Download Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) - [Download Nvidia® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) -Install Geti Windows application and launch it from the Start menu +Install Geti Windows application and launch it from the Start menu. ### Docker @@ -67,7 +72,7 @@ docker pull ghcr.io/open-edge-platform/geti-xpu # modern Intel® CPU/GPU (rec docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® CUDA platforms docker pull ghcr.io/open-edge-platform/geti-cpu # CPU-only (most lightweight) -# Retag the pulled image as `geti-{cpu,xpu,cuda}:latest` for using with `just run-image` +# Retag the pulled image as `geti-{cpu,xpu,cuda}:latest` for use with `just run-image` docker tag ghcr.io/open-edge-platform/geti-cpu:latest geti-cpu:latest just run-image --accelerator xpu # launch the application @@ -78,9 +83,9 @@ Then open the Geti web application at [**http://localhost:7860**](http://localho For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the [application README](application/README.md). -#### Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export +### Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export) -Linux, WSL (In order to run a script you need to have curl & git installed): +Linux, WSL (in order to run the script you need to have curl & git installed): ```bash curl -fsSL https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.sh | bash @@ -124,7 +129,6 @@ exported_path = engine.export() # writes OpenVINO IR See the [library README](library/README.md) for the full list of recipes, advanced configuration, dataset support, and inference/optimization examples. - ## Key Features
@@ -132,12 +136,31 @@ inference/optimization examples. 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. +

+ Application demo +

+ +
+ +
+⚡ Hardware-accelerated inference & model optimization + +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. + +
+ +
+🚀 Integrated deployment & inference + +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. +
🏆 State-of-the-art model catalog -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)! +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)! @@ -235,56 +258,24 @@ Train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, Y
🎨 Multiple computer vision tasks -Image classification, object detection, and instance segmentation from the no-code web interface, with even more tasks available through the Python API (`getitune`). +Geti™ supports multiple computer vision 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.
🧠 Smart annotations -Manual and semi-automated labeling powered by models like SAM (Segment Anything Model), plus bulk labeling to dramatically speed up dataset creation. - -
- -
-📦 Dataset & model versioning - -Track how datasets and models evolve, link models to a specific dataset revision, view exact training hyperparameters, and fine-tune from any previous version. - -
- -
-🏔️ Runs locally, on the edge - -Fine-tune models 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. Minimum recommended setup: **8 CPU threads, 16 GB RAM, 40 GB free disk**. - -
- -
-⚡ Hardware acceleration - -Optimized for modern Intel® hardware (Arc™ GPUs, Core™ Ultra processors). 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; NVIDIA® CUDA and CPU-only execution are also supported. - -
- -
-💾 Dataset import & export - -COCO, Pascal VOC, and YOLO formats plus a Geti-optimized native format, with label filtering to selectively include or exclude labels on import/export. - -
- -
-🔧 Model optimization - -Built-in quantization with accuracy-aware INT8 optimization to balance inference speed and accuracy on resource-constrained edge devices. +Smart annotations in Geti™ enable users to easily create bounding boxes, rotated bounding boxes, segmentation boundaries, and more. 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. +

+ Smart Annotations +

-🚀 Integrated deployment & inference +📦 Model & dataset management -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. +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.
@@ -310,7 +301,6 @@ Geti™ is a powerful tool to build vision models for a wide range of processes, - [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/) - ## Community - To report a bug or submit a feature request, please open a [GitHub issue](https://github.com/open-edge-platform/geti/issues). From 0e7e81f1f98382844dff65206d9cd4bc80129599 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 16:05:11 +0100 Subject: [PATCH 06/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index bd7af9b538..43877e8a96 100644 --- a/README.md +++ b/README.md @@ -145,7 +145,7 @@ Geti™ enables users to start building deep-learning computer vision models wit
⚡ Hardware-accelerated inference & model optimization -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. +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.
@@ -282,12 +282,12 @@ Track how datasets and models evolve, link models to a specific dataset revision ## Ecosystem - [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) → visual prompting -- [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) → robot learning, VLA (Vision-Language-Action) +- [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 preprocess and postprocess as well as routine procedures. +- [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) - An nd-to-end framework for teaching robots to perform tasks through imitation learning from human demonstrations. ## Who uses Geti™? From 0754d470176ab2587fa3b97effa164690d8ccc82 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 16:36:18 +0100 Subject: [PATCH 07/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 46 ++++++++++++++++++++++++++++++---------------- 1 file changed, 30 insertions(+), 16 deletions(-) diff --git a/README.md b/README.md index 43877e8a96..8303048bdd 100644 --- a/README.md +++ b/README.md @@ -35,19 +35,19 @@ Geti™ is an end-to-end Vision AI application that takes you from raw images to > but also to receive better support from Intel and the Geti™ community. > For upgrade from Geti™ v2 to v3, please follow the [upgrade guidance](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/migration-from-geti-2x). -# Quick Start +## Quick start with Geti™ Get Geti running and train your first model in a few minutes. For full instructions and all options, see the [official documentation](https://docs.geti.intel.com/) and the [application README](application/README.md). -**Minimum recommended setup:** 8 CPU threads, 16 GB RAM, 40 GB free disk. A GPU (Intel® XPU or NVIDIA® CUDA) is -recommended for larger models. +**Minimum recommended setup** -| Component | Minimum requirement | -| --------- | ------------------- | -| CPU | 8 threads | -| RAM | 16 GB | -| Disk | 40 GB free | +| Component | Requirement | +| --------- | ------------------------------------------------------- | +| CPU | 8 threads | +| RAM | 16 GB | +| Disk | 40 GB free | +| GPU | Optional — Intel® XPU or NVIDIA® CUDA for larger models | ## 1. Run Geti @@ -63,7 +63,7 @@ Download the Windows Installer: Install Geti Windows application and launch it from the Start menu. -### Docker +### Container image Pull a pre-built image for your hardware and launch it: @@ -83,7 +83,9 @@ Then open the Geti web application at [**http://localhost:7860**](http://localho For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the [application README](application/README.md). -### Install natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export) +### Install from souurce code + +natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export) Linux, WSL (in order to run the script you need to have curl & git installed): @@ -104,15 +106,26 @@ Once Geti is running, build your first model directly in the web UI: See [Training your first model](https://docs.geti.intel.com/) for the full walkthrough. -### Use the Python API (`getitune`) +## Quick start with `getitune` + +The Geti™ training engine is published on PyPI and can train, optimize, and deploy models. -Prefer to work programmatically? Geti's training engine is published on PyPI and can train, optimize, and deploy models -from Python. It requires **Python 3.11–3.14**, **PyTorch 2.10**, **OpenVINO™ 2026.1**, and **NumPy ≥ 2.0**. +To install `getitune`: ```bash -pip install "getitune[cpu]" # or [xpu] for Intel® GPU, [cuda] for NVIDIA® GPU +# With uv (recommended) +uv pip install "getitune" + +# Or with pip +pip install "getitune" ``` +> [!NOTE] +> For advanced installation options, including hardware-specific PyTorch wheels, +> see the [`getitune` README](library/README.md). + +Provide `getitune` with a dataset and fine-tune a model: + ```python from getitune.engine import create_engine @@ -126,8 +139,9 @@ engine.test() exported_path = engine.export() # writes OpenVINO IR ``` -See the [library README](library/README.md) for the full list of recipes, advanced configuration, dataset support, and -inference/optimization examples. +> [!NOTE] +> See the [`getitune` README](library/README.md) for the full list of recipes, advanced configuration, dataset support +> and inference/optimization examples. ## Key Features From fb6c144a1a2a517a81a2a31929c90ba542f836e8 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 17:25:36 +0100 Subject: [PATCH 08/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 55 ++++++++++++++++++++++++++++--------------------------- 1 file changed, 28 insertions(+), 27 deletions(-) diff --git a/README.md b/README.md index 8303048bdd..1da7e09e4a 100644 --- a/README.md +++ b/README.md @@ -37,21 +37,19 @@ Geti™ is an end-to-end Vision AI application that takes you from raw images to ## Quick start with Geti™ -Get Geti running and train your first model in a few minutes. For full instructions and all options, see the -[official documentation](https://docs.geti.intel.com/) and the [application README](application/README.md). - -**Minimum recommended setup** +Before you begin, make sure your machine meets the following requirements: | Component | Requirement | | --------- | ------------------------------------------------------- | | CPU | 8 threads | | RAM | 16 GB | | Disk | 40 GB free | -| GPU | Optional — Intel® XPU or NVIDIA® CUDA for larger models | +| GPU | Optional — Intel® XPU or NVIDIA® GPU for larger models | -## 1. Run Geti +Geti can be installed as a **Windows application**, run as a **container**, or built **from source code**. Choose the option that best suits your environment below. -### Windows Application +
+Windows Application Run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments. @@ -59,11 +57,14 @@ Download the Windows Installer: - [Download CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) - [Download Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) -- [Download Nvidia® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) +- [Download NVIDIA® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) Install Geti Windows application and launch it from the Start menu. -### Container image +
+ +
+Container image Pull a pre-built image for your hardware and launch it: @@ -83,32 +84,36 @@ Then open the Geti web application at [**http://localhost:7860**](http://localho For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the [application README](application/README.md). -### Install from souurce code +
+ +
+Install from source code -natively with Ultralytics YOLO26 models (the latest NMS‑free, edge‑optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export) +Install Geti natively on Linux or WSL, including support for Ultralytics YOLO26 models — the latest NMS-free, edge-optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export. -Linux, WSL (in order to run the script you need to have curl & git installed): +Requires `curl` and `git`. Run the following on Linux or WSL: ```bash curl -fsSL https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.sh | bash ``` -## 2. Train your first model +
-Once Geti is running, build your first model directly in the web UI: +Once Geti is up and running, follow the intuitive UI to train your first model. -1. **Create a project** — choose a task (object detection, instance segmentation, or classification) and define your labels. -2. **Upload media** — drag in 20–50 representative images to start. -3. **Annotate** — label your media with the built-in manual and AI-assisted tools. -4. **Train** — pick a recommended architecture and start training; watch progress in the Jobs panel. -5. **Deploy** — build an inference pipeline (source → model → sink) and run predictions in real time, or export an - OpenVINO™-optimized bundle for the edge. +

+ Application demo +

-See [Training your first model](https://docs.geti.intel.com/) for the full walkthrough. +> [!NOTE] +> See the detailed step-by-step guidance on how to train your first model in +> ["Training your first model"](https://docs.geti.intel.com/docs/user-guide/quick-start/training-your-first-model) +> section in the Geti™ documentation. +> Full instructions and all options are available in [Geti™ documentation](https://docs.geti.intel.com/). ## Quick start with `getitune` -The Geti™ training engine is published on PyPI and can train, optimize, and deploy models. +The Geti™ training engine `getitune` is published on PyPI and can train, optimize, and deploy models. To install `getitune`: @@ -148,11 +153,7 @@ exported_path = engine.export() # writes OpenVINO IR
🔄 Interactive end-to-end model training -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. - -

- Application demo -

+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.
From 5f8d6566620be7adbfb1adbfe921e8d5e771f3ec Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 18:06:53 +0100 Subject: [PATCH 09/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 70 +++++++++++++++++++++++++++---------------------------- 1 file changed, 35 insertions(+), 35 deletions(-) diff --git a/README.md b/README.md index 1da7e09e4a..25f00ddb7b 100644 --- a/README.md +++ b/README.md @@ -39,64 +39,67 @@ Geti™ is an end-to-end Vision AI application that takes you from raw images to Before you begin, make sure your machine meets the following requirements: -| Component | Requirement | -| --------- | ------------------------------------------------------- | -| CPU | 8 threads | -| RAM | 16 GB | -| Disk | 40 GB free | -| GPU | Optional — Intel® XPU or NVIDIA® GPU for larger models | +| Component | Requirement | +| --------- | ------------------------------------------------------ | +| CPU | 8 threads | +| RAM | 16 GB | +| Disk | 40 GB free | +| GPU | Optional - Intel® XPU or NVIDIA® GPU for larger models | Geti can be installed as a **Windows application**, run as a **container**, or built **from source code**. Choose the option that best suits your environment below.
Windows Application -Run Geti as a native Windows application, with prebuilt images for Intel® XPU, NVIDIA® CUDA, and CPU-only environments. +Download the latest Geti™ Windows installer suitable for your hardware (Intel® XPU, NVIDIA® CUDA or CPU-only) from the [releases repository](https://storage.geti.intel.com/geti/packages): -Download the Windows Installer: +- [CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) +- [Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) +- [NVIDIA® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) -- [Download CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) -- [Download Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) -- [Download NVIDIA® CUDA version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cuda-3.0.0.msix) - -Install Geti Windows application and launch it from the Start menu. +Install Geti™ Windows application and launch it from the Start menu.
Container image -Pull a pre-built image for your hardware and launch it: +Pull a pre-built container image for your hardware and launch it using [`just`](https://just.systems), which handles device passthrough, volumes, and WebRTC ports automatically: ```bash +# 1. Install just +curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s -- --to /usr/local/bin + +# 2. Clone the repository +git clone https://github.com/open-edge-platform/geti.git +cd geti/application + +# 3. Pull the image for your hardware docker pull ghcr.io/open-edge-platform/geti-xpu # modern Intel® CPU/GPU (recommended) -docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® CUDA platforms +docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® GPU (CUDA) docker pull ghcr.io/open-edge-platform/geti-cpu # CPU-only (most lightweight) -# Retag the pulled image as `geti-{cpu,xpu,cuda}:latest` for use with `just run-image` -docker tag ghcr.io/open-edge-platform/geti-cpu:latest geti-cpu:latest +# 4. Retag the pulled image for use with just +docker tag ghcr.io/open-edge-platform/geti-xpu:latest geti-xpu:latest -just run-image --accelerator xpu # launch the application +# 5. Launch the application +just run-image --accelerator xpu ``` -Then open the Geti web application at [**http://localhost:7860**](http://localhost:7860). - -For build-from-source options and advanced setup, see the [installation guide](https://docs.geti.intel.com/) and the -[application README](application/README.md). +Then get access to Geti™ user interface at `http://localhost:7860`.
Install from source code - -Install Geti natively on Linux or WSL, including support for Ultralytics YOLO26 models — the latest NMS-free, edge-optimized models (Nano / Small / Medium) for object detection and instance segmentation. The integration covers the full model lifecycle: training, inference, quantization, and OpenVINO™ model export. - -Requires `curl` and `git`. Run the following on Linux or WSL: +To install the Geti™ stable development version from source code, use: ```bash curl -fsSL https://raw.githubusercontent.com/open-edge-platform/geti/develop/install.sh | bash ``` +Installing from source gives you access to the latest features not yet available in released builds, including Ultralytics YOLO26 support. +
Once Geti is up and running, follow the intuitive UI to train your first model. @@ -107,9 +110,10 @@ Once Geti is up and running, follow the intuitive UI to train your first model. > [!NOTE] > See the detailed step-by-step guidance on how to train your first model in -> ["Training your first model"](https://docs.geti.intel.com/docs/user-guide/quick-start/training-your-first-model) -> section in the Geti™ documentation. +> ["Training your first model"](https://docs.geti.intel.com/docs/user-guide/quick-start/training-your-first-model) section in the Geti™ documentation. > Full instructions and all options are available in [Geti™ documentation](https://docs.geti.intel.com/). +> +> Detailed installation guide is available in ["Installation guide"](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/installation-guide) ## Quick start with `getitune` @@ -125,10 +129,6 @@ uv pip install "getitune" pip install "getitune" ``` -> [!NOTE] -> For advanced installation options, including hardware-specific PyTorch wheels, -> see the [`getitune` README](library/README.md). - Provide `getitune` with a dataset and fine-tune a model: ```python @@ -145,8 +145,8 @@ exported_path = engine.export() # writes OpenVINO IR ``` > [!NOTE] -> See the [`getitune` README](library/README.md) for the full list of recipes, advanced configuration, dataset support -> and inference/optimization examples. +> See the [`getitune` README](library/README.md) for the full list of recipes, advanced configuration, dataset support, +> inference/optimization examples, and hardware-specific PyTorch installation options. ## Key Features @@ -302,7 +302,7 @@ Track how datasets and models evolve, link models to a specific dataset revision - [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 preprocess and postprocess as well as routine procedures. -- [Physical AI Studio](https://github.com/open-edge-platform/physical-ai-studio) - An nd-to-end framework for teaching robots to perform tasks through imitation learning from human demonstrations. +- [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™? From ecd9ccb501634122959d1cadd310ecec85c5acda Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 18:13:00 +0100 Subject: [PATCH 10/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index 25f00ddb7b..74988f031d 100644 --- a/README.md +++ b/README.md @@ -33,7 +33,7 @@ Geti™ is an end-to-end Vision AI application that takes you from raw images to > Previous versions of Geti™ are still available in a separate [repository](https://github.com/open-edge-platform/geti_v2). > In general, we recommend upgrading to the latest Geti™ release whenever possible - not only to access new functionality, > but also to receive better support from Intel and the Geti™ community. -> For upgrade from Geti™ v2 to v3, please follow the [upgrade guidance](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/migration-from-geti-2x). +> To upgrade from Geti™ v2 to v3, please follow the [upgrade guidance](https://docs.geti.intel.com/docs/user-guide/getting-started/installation/migration-from-geti-2x). ## Quick start with Geti™ @@ -273,7 +273,7 @@ Would you like to see a specific model added? Let us know by opening a [GitHub i
🎨 Multiple computer vision tasks -Geti™ supports multiple computer vision 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. +Geti™ supports multiple computer vision 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.
@@ -301,7 +301,7 @@ Track how datasets and models evolve, link models to a specific dataset revision - [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 preprocess and postprocess as well as routine procedures. +- [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™? From dac7557a97b3194fda4c4b52b046e14c639933c6 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Fri, 19 Jun 2026 18:15:14 +0100 Subject: [PATCH 11/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 74988f031d..d2ea2cd2d0 100644 --- a/README.md +++ b/README.md @@ -51,7 +51,7 @@ Geti can be installed as a **Windows application**, run as a **container**, or b
Windows Application -Download the latest Geti™ Windows installer suitable for your hardware (Intel® XPU, NVIDIA® CUDA or CPU-only) from the [releases repository](https://storage.geti.intel.com/geti/packages): +Download the latest Geti™ Windows installer suitable for your hardware (Intel® XPU, NVIDIA® CUDA or CPU-only) from the releases repository: - [CPU-only version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-cpu-3.0.0.msix) - [Intel® XPU version installer](https://storage.geti.intel.com/geti/packages/3.0.0/geti-xpu-3.0.0.msix) From e41b4f4fb97ebb0c133d8f89fb863fdc7fbbc2c0 Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Mon, 22 Jun 2026 08:47:40 +0100 Subject: [PATCH 12/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 50 +++++++++++++++++++++++++------------------------- 1 file changed, 25 insertions(+), 25 deletions(-) diff --git a/README.md b/README.md index d2ea2cd2d0..6ab65f8553 100644 --- a/README.md +++ b/README.md @@ -150,28 +150,7 @@ exported_path = engine.export() # writes OpenVINO IR ## Key Features -
-🔄 Interactive end-to-end model training - -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. - -
- -
-⚡ Hardware-accelerated inference & model optimization - -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. - -
- -
-🚀 Integrated deployment & inference - -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. - -
- -
+
🏆 State-of-the-art model catalog Train and fine-tune modern architectures such as RF-DETR, DINOv3 DETR, YOLO26, YOLOX, D-FINE, and Mask R-CNN. @@ -270,6 +249,27 @@ Would you like to see a specific model added? Let us know by opening a [GitHub i
+
+🔄 Interactive end-to-end model training + +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. + +
+ +
+⚡ Hardware-accelerated inference & model optimization + +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. + +
+ +
+🚀 Integrated deployment & inference + +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. + +
+
🎨 Multiple computer vision tasks @@ -335,10 +335,10 @@ For those who would like to contribute, see [Contributing guide](CONTRIBUTING.md ## License -Geti™ is licensed under the [Apache License Version 2.0](LICENSE). By contributing to the project, you agree to the -license and copyright terms therein and release your contribution under these terms. -Stay tuned for further updates soon! +Geti™ is licensed under the [Apache License Version 2.0](LICENSE). ## Disclaimers +Geti™ utilizes FFmpeg. + FFmpeg is an open source project licensed under LGPL and GPL. See [https://www.ffmpeg.org/legal.html](https://www.ffmpeg.org/legal.html). You are solely responsible for determining if your use of FFmpeg requires any additional licenses. Intel is not responsible for obtaining any such licenses, nor liable for any licensing fees due, in connection with your use of FFmpeg. From 3bee1af2c3164cbd81a5ebb4c00bbfa8c75db9ad Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Mon, 22 Jun 2026 08:53:50 +0100 Subject: [PATCH 13/17] readme Signed-off-by: Barabanov, Alexander --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 6ab65f8553..93f6a4681e 100644 --- a/README.md +++ b/README.md @@ -308,6 +308,7 @@ Track how datasets and models evolve, link models to a specific dataset revision Geti™ is a powerful tool to build vision models for a wide range of processes, including detecting defective parts in a production line, reducing downtime on the factory floor, automating inventory management, or other automation projects. We have chosen to highlight a few interesting community members: +- [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/) From dda229ebb3bbdad0f91206ef1196cd8612b474f0 Mon Sep 17 00:00:00 2001 From: Alexander Barabanov <97449232+AlexanderBarabanov@users.noreply.github.com> Date: Mon, 22 Jun 2026 09:44:04 +0100 Subject: [PATCH 14/17] Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 93f6a4681e..b6e26791a2 100644 --- a/README.md +++ b/README.md @@ -331,7 +331,7 @@ For those who would like to contribute, see [Contributing guide](CONTRIBUTING.md

- Contributors + Contributors ## License From a0c09ab28ef341d73dadff41d30df1c594891451 Mon Sep 17 00:00:00 2001 From: Alexander Barabanov <97449232+AlexanderBarabanov@users.noreply.github.com> Date: Mon, 22 Jun 2026 09:44:13 +0100 Subject: [PATCH 15/17] Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index b6e26791a2..1da50b6c9e 100644 --- a/README.md +++ b/README.md @@ -273,7 +273,7 @@ Build custom pipelines (source → model → sink) to deploy models inside Geti
🎨 Multiple computer vision tasks -Geti™ supports multiple computer vision 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. +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.
From dd2c0f6b2c32e0d00161a8398efec53fe9c781b7 Mon Sep 17 00:00:00 2001 From: Alexander Barabanov <97449232+AlexanderBarabanov@users.noreply.github.com> Date: Mon, 22 Jun 2026 09:44:23 +0100 Subject: [PATCH 16/17] Potential fix for pull request finding Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com> --- README.md | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) diff --git a/README.md b/README.md index 1da50b6c9e..f166c84009 100644 --- a/README.md +++ b/README.md @@ -74,16 +74,15 @@ curl --proto '=https' --tlsv1.2 -sSf https://just.systems/install.sh | bash -s - git clone https://github.com/open-edge-platform/geti.git cd geti/application -# 3. Pull the image for your hardware -docker pull ghcr.io/open-edge-platform/geti-xpu # modern Intel® CPU/GPU (recommended) -docker pull ghcr.io/open-edge-platform/geti-cuda # NVIDIA® GPU (CUDA) -docker pull ghcr.io/open-edge-platform/geti-cpu # CPU-only (most lightweight) +# 3. Pull the image for your hardware (choose one: cpu|cuda|xpu) +ACCELERATOR=xpu +docker pull ghcr.io/open-edge-platform/geti-${ACCELERATOR} # 4. Retag the pulled image for use with just -docker tag ghcr.io/open-edge-platform/geti-xpu:latest geti-xpu:latest +docker tag ghcr.io/open-edge-platform/geti-${ACCELERATOR}:latest geti-${ACCELERATOR}:latest # 5. Launch the application -just run-image --accelerator xpu +just run-image --accelerator ${ACCELERATOR} ``` Then get access to Geti™ user interface at `http://localhost:7860`. From 6ca30f25de96a79a25ddffbfac8b8e0630fc37ec Mon Sep 17 00:00:00 2001 From: "Barabanov, Alexander" Date: Mon, 22 Jun 2026 10:53:59 +0100 Subject: [PATCH 17/17] fix review Signed-off-by: Barabanov, Alexander --- README.md | 34 +++++++++++++++++++++++++++------- 1 file changed, 27 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index f166c84009..de7d36465c 100644 --- a/README.md +++ b/README.md @@ -5,7 +5,7 @@ [Quick Start](#quick-start) • [Geti™ documentation](https://docs.geti.intel.com/) • -[`getitune` documentation](https://open-edge-platform.github.io/geti/latest/index.html) +[`getitune` documentation](library/README.md) [![Container build](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml/badge.svg)](https://github.com/open-edge-platform/geti/actions/workflows/build.yaml) [![Codecov](https://codecov.io/gh/open-edge-platform/geti/branch/develop/graph/badge.svg?token=9HVFNMPFGD)](https://codecov.io/gh/open-edge-platform/geti) @@ -130,17 +130,37 @@ pip install "getitune" Provide `getitune` with a dataset and fine-tune a model: -```python +```Python +from getitune.utils import list_models from getitune.engine import create_engine +from getitune.types import ExportFormat, ExportPrecision -# Initialize and train using a bundled recipe and dataset +# List all available models names +all_models = list_models() + +# create Engine engine = create_engine( - data="tests/assets/classification_cifar10", - model="src/getitune/recipe/classification/multi_class_cls/efficientnet_b0.yaml", + model="efficientnet_b0", + data="/path/to/dataset", + work_dir="./my_workspace", ) + +# train a model engine.train() -engine.test() -exported_path = engine.export() # writes OpenVINO IR + +# Export to FP32 OpenVINO IR (default) +ov_ir_path = engine.export() + +# validate engine +ov_engine = create_engine( + model="/path/to/exported_model.xml", + data="/path/to/dataset", +) +ov_engine.test() # test on test subset +ov_engine.predict() # predict on test subset + +# optimize a model to int8 quantized version via NNCF tool +ov_engine.optimize() ``` > [!NOTE]
Computer Vision TaskComputer Vision Task Model Architecture Paper