v2.13.1

Release Notes

Geti keeps evolving - and thanks to your feedback, version 2.13.1 contains correction of several important issues.

Summary of major features and improvements

• Setting the constant name of the k3s node - which improves behavior of the Geti when host name is changed on an environment - detailed issue description
• Choosing an Intel device driver during installation has been improved - which resolves the issue described here
• Live prediction for MaskRCNN models has been corrected - the issue describing this problem is here
• Several smaller issues related to upgrade and revert have been corrected as well.

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

Source code for open source components, provided as required by license.

v2.13.0

Release Notes

Geti evolution continues with version 2.13, bringing enhanced AI capabilities, improved collaboration options, and smarter workflows to help you scale computer vision projects faster and more efficiently.

Summary of major features and improvements

• Model Fine-Tuning Support on Intel® Core™ Ultra (Series 2): Geti now supports computer vision model fine-tuning on Intel® Core™ Ultra (Series 2) processors with integrated GPU (iGPU).
• New SOTA Detection Algorithm – DEIM: Introducing DEIM-DETR with Dense One-to-One Matching, delivering significant accuracy gains, especially for small object detection (+1.5 AP vs. D-FINE-X). Geti now supports DEIM as the latest state-of-the-art model for object detection tasks.
• Configurable Data Augmentation for All Tasks: Gain full control over how your training data is augmented. Users can now configure augmentation parameters across all tasks enhancing both control and model robustness.
• Support for Annotation Shapes with Holes: Users can now mark regions inside annotations as holes for semantic segmentation tasks. This functionality will help users to better understand the inferred object shape and dimensions, for example as part of manufacturing use cases or medical imaging.
• Multiple Workspaces Support: Users can now separate teams within a single Geti instance using multiple workspaces, ensuring data governance and privacy while sharing the same compute resources. Flexible, role-based access control is now available not only at the organization and project levels, but also at the workspace level.
• Selective Model Export: Reduce project export size and export time by choosing which models to include in your project archive. Export all models, only the latest active one, or none — depending on your needs.
• Additional Hotkey for Faster Annotation: In case of images that do not contain any regions of interest, users can now mark those with a single click as "No Object" (object detection) or "Empty" (segmentation) using a dedicated hotkey to further accelerate data annotation.

Known issues

• If you are installing Geti on a system with Arrow Lake CPU and no discrete GPU, you will face an issue described here. Follow the workaround from the ticket to be able to fine tune your models.

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

Source code for open source components, provided as required by license.

v2.12.1

Release Notes

Geti keeps evolving - and thanks to your feedback, version 2.12.1 contains correction of one important error.

Summary of major features and improvements

• Fixed the handling of projects that were registered in v2.11.0 after upgrading to v2.12.0 - detailed issue description

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

Source code for open source components, provided as required by license.

v2.12.0

Release Notes

Geti keeps evolving - and thanks to your feedback, version 2.12.0 introduces several improvements and new features to elevate your experience.

Summary of major features and improvements

• Gold Support for Model Fine-Tuning on Intel Arc GPU: Succeeding the Beta Support from Geti™ 2.10.0 pre-release, you can now fine-tune models across the entire Geti model suite with performance on par with top-tier GPUs. Any Arc card will be supported (BMG, Alchemist), but we recommend using cards with 16GB VRAM or more. Even though 12GB cards will also be supported, a few large models may encounter issues during training.

• Streamlined Advanced Training Settings: The model training interface has been completely updated with a more intuitive and cohesive design to select the model architecture and configure the various advanced training settings. In addition, new advanced parameters have been exposed:

  • Configurable data augmentation as part of ‘Data Management’ for classification tasks (will be applied to the other Geti tasks in upcoming releases), so that users can adjust the augmentation settings based on their needs;
  • Configurable model input size as part of ‘Training’, so that users can control the resolution at which images are processed for model fine-tuning.

• Improved Media Upload Reliability: Various improvements to the media upload logic to make it faster and more reliable, including a better algorithm to detect corrupted videos and attempt repairing them on the fly.

• Added support for key point detection dataset import and export in Datumaro format.

Changes to the REST API

For more details, we recommend you refer to the full API specification.

• Added new endpoints to get and update the project and training configuration.
• Deprecated the legacy configuration endpoints.
• Updated the response of the supported algorithms endpoint.
• Several fields have been added, removed or renamed.
• Added a new query parameter to the project export endpoint to control which models to include in the exported zip archive (choice between all models, only the active one or none).
• Removed a deprecated set of endpoints to download the models along with sample scripts (aka code deployment).

Model deprecations

The following model architectures are now deprecated. While it is still possible to use them in 2.12, users are recommended to switch to one of the newer, higher-performance model architectures.

• Detection
  • ATSS-ResNeXt101
  • RTDetr-R18
  • RTDetr-R101
  • RTMDet-Tiny
• Rotated detection
  • MaskRCNN-ResNet50-V1
• Classification
  • EfficientNet-V2-L
  • MobileNet-V3-small
• Instance Segmentation:
  • MaskRCNN-ResNet50-V1
• Semantic Segmentation
  • LiteHRNet-X

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

Source code for open source components, provided as required by license.

v2.11.0

Release Notes

Geti keeps evolving - and thanks to your feedback, version 2.11.0 introduces several improvements and new features to elevate your experience.

Summary of major features and improvements

• 💡 Added Support for Single Object Keypoint Detection: As part of Geti 2.11, this new task allows users to identify and locate specific points of interest (keypoints) in an image, enabling use cases such as pose estimation. With its custom annotation tool, users can prepare their own keypoint detection datasets to train their model.

• 🖥️ Optimized Resource Utilization for Lower-Spec Hardware: Geti can now run on systems with 16 CPU cores and 32 GB RAM. For users who have large datasets or want to use heavy models, 64 GB of memory is still recommended.

• 🔄 Platform Upgrades Now Supported via Helm Charts: This facilitates a direct upgrade of your Geti instance without requiring the use of the installer.

• 🚀 Increased Inference Efficiency: FP16 models generate predictions with an accuracy comparable to their FP32 version, but with significantly reduced latency and memory footprint. By defaulting to FP16, Geti inference is now more responsive and resource efficient.

• ☁️ Installation Guide for AWS & Azure cloud VM: A comprehensive, step-by-step guide is now available for installing and configuring Intel Geti on AWS and Azure. It covers the virtual machines setup, recommended configurations, installation, upgrades, and best practices for maintaining Geti in the cloud – designed to be clear and accessible for everyone.

• ☁️ Geti™ Available on AWS Marketplace: Geti can now be deployed via AWS Marketplace for free, making it more accessible to users who are already operating in the AWS cloud

• 🎨 Interface and Workflow Improvements:

  • Job filtering by time range: to make it easier for users to keep track on their work, the new calendar-based filter can be used to show jobs started within a given time range.

  • Training job visibility: for each project, all scheduled and in-progress training jobs are now listed on the Models screen for better visibility.

  • Label ordering: users can now rearrange the order of labels and their groups as they like, so that they can put the labels that they are most interested in on top for better visibility.

  • Project import and renaming: it is now possible to directly assign a new name to a project before uploading it to avoid project name duplication.

  • Improved user flow for Live Prediction: as part of Geti’s testing capabilities, we improved the user experience of Live Prediction. With minimal clicks, users can now directly use their camera inside the Tests screen to capture a test image.

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

Source code for open source components, provided as required by license.

v2.10.2

Release Notes

Bugfixes

• Fixed a bug (#486) where Geti installation failed because bitnami-shell container image was removed from the Bitnami repository (PR #513)

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

v2.10.1

Release Notes

Summary of major features and improvements

• YOLO-X is used as a default model for object detection training on hardware setups with Intel GPU in #282 and #260
• Memory limits on weight uploader and user directory was increased in #306 to prevent OOM occurrences during installation
• Relaxed the OS version check in the installer so that users can install Intel Geti on untested Ubuntu versions

Bugfixes

• Fixed UI issues related to loading test results in #300
• Solved issue with Intel® Arc A-Series™ discovery in installer

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

v2.10.0

Release Notes

Now that the complete Intel® Geti™ codebase is available on GitHub, we will continue to improve our software to meet your needs. Following the early feedback, we received after publishing Intel® Geti™ v2.9.0, we accommodated improvements in our new 2.10.0 release to make it easier for you to install the software on your hardware setup of choice. Check out the updates below!

Summary of major features and improvements

• 🚀 Beta Support for Model Fine-Tuning on Intel Arc GPU: With the new Intel® Geti™ 2.10.0 pre-release, we now provide Beta support to run model training on the Intel Arc GPU. Any Arc card will be supported (BMG, Alchemist), but we recommend using cards with 16GB VRAM or more. Even though 12GB cards will also be supported, a few large models may encounter issues during training. We will continue to work on Gold support for Intel Arc GPU to further optimize the model performance across the whole Intel® Geti™ model suite. With this pre-release, you can start to experience how Intel hardware enables you to build AI in minimum amount of time!

• 🔓 Increased Installation Flexibility: To ensure that the recommended hardware requirements are not a blocker for you to install and use Intel® Geti™, we made the hardware validation during installation more flexible. This means that you can now easily install Intel® Geti™ on lower-spec systems:

  • Systems with GPUs that have less than 16GB of VRAM;
  • Systems that have less than 64GB of OS memory;
  • Systems with 16 OS cores at minimum;
  • Systems with smaller disk space than 500GB, with 100GB at minimum;
  • Systems without GPU. If no GPU is present, model training will be run on the CPU. However, for the best model training performance, we recommend using systems with a dedicated GPU.

The installer will inform you if any of our recommended hardware requirements are not met, ensuring you know how to achieve the best experience with our software as a few large models may encounter issues during training on lower-spec hardware. However, this will not stop you from successfully installing Intel® Geti™ and exploring its capabilities.

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*

v2.9.0

Release Notes

We are honored to share that the complete Intel® Geti™ codebase is now available on GitHub! Join us on this exciting journey as Intel® Geti™ opens its doors to the community. Dive into the code, contribute, and innovate with us!

Summary of major features and improvements

• 🔍State-of-the-Art Detection Model: Introducing the D-fine detection model, a breakthrough in object detection technology. Now integrated into Intel® Geti™, it's ready for fine-tuning to meet your specific needs.

• 🌀U-flow Anomaly Detection: Elevate your anomaly detection workflows with the U-flow distribution map-based model, now part of Intel® Geti™. Explore this advanced training option for unparalleled anomaly detection capabilities.

• 🧩Interactive OpenAPI documentation: Discover and explore Intel® Geti™ API endpoints directly within our user interface, thanks to the integration of the Scalar reference API component. This feature allows you to experiment with API requests in your browser, making it easier than ever to understand and utilize our API capabilities. Access the OpenAPI specification by visiting /rest-api/openapi-specification.

• 🌟Seamless Helm Chart Installation: Deploy Intel® Geti™ effortlessly on any Kubernetes distribution, whether on-premises or in the cloud, using our official Helm chart. Follow our step-by-step installation guide to get started here.

• 🐳Public Container Images: Access official Intel® Geti™ container images directly from the GitHub Container Registry. It's never been easier to integrate Intel® Geti™ into your workflows!

• ⚡Optimized Installer: Experience a leaner, meaner Intel® Geti™ installer! We've refactored the installer to utilize public container images, resulting in a significantly smaller and more efficient package. Check out the details here.

Download this release

Installer

Download installer.

Container Images and Helm charts

Container images and Helm charts are available at ghcr.io/open-edge-platform/geti/*