Intel® Geti™ 1.8.0#

Release Summary#

Intel® Geti™ 1.8.0 contains several updates and feature enhancements, including key highlights:

Other Intel® Geti™ 1.8.0 updates include:

Release Details#

This section covers additional details on the new functionality available with Intel® Geti™ 1.8.0.

Enhanced labeling experience with the Automatic Segmentation tool#

Leverage the new Automatic Segment annotation tool to segment objects and create bounding boxes with a click. This new functionality is available in the toolbar of the Annotation screen.

Hover over any object in an image or frame to preview the annotation, select the object to apply the mark, and fine-tune the results by selecting areas to add or remove.

Placeholder

Read the documentation on this new functionality to learn more.

Sample datasets available#

Intel® Geti™ 1.8.0 includes access to sample datasets available through the User Guide. These datasets can help users familiarize themselves with the platform and create models with different vision tasks, such as object detection and semantic segmentation.

Note

The usage of these datasets is reserved only for getting started with the platform, any commercial uses are not allowed.

New storage tab#

The storage tab under the Account menu features visibility into total storage usage.

Placeholder

Video player improvements#

The Intel® Geti™ platform 1.8.0 brings along a suite of enhancements aimed at improving the functionality and user experience of the video player.

Key improvements include:

  • The video player timeline will accurately show predictions and annotations during the video playback, sometimes requiring a buffer time for a seamless viewing experience.

  • During playback of video files, if the playback speed is adjusted, the video will seamlessly continue playing at the new speed.

  • In case a video playback reaches the end, clicking the ‘play’ button will restart the video from the beginning.

  • After the completion of a video, the frame marker will now align with the last frame.

Project size display#

Project size information is now available in the Projects tab by selecting the details view.

Placeholder

Removal of Filter Pruning#

The Filter Pruning function that provided advanced model optimization capability from the platform has been removed.

This decision was based on several factors:

  • Filter Pruning requires large datasets.

  • It is difficult to use for modern models, e.g., Vision Transformers.

  • We have observed low adoption rates of Filter Pruning, even outside of the Intel® Geti™ platform.

This change aligns with our commitment to providing users with the most efficient and relevant tools for model optimization.

Download individual media#

Users can now download specific images or videos available in the datasets directly via the platform’s UI. They can select one or multiple media files for export. This addition is particularly useful for users capturing new images for training via an external data source, connected with the Intel® Geti™ platform via the API.

This function can be performed from both the Dataset screen and the Annotator screen.

Placeholder

Active model architecture indication#

The active model architecture and version information are now available in the auto-training status area on the Annotator screen. This information can help users avoid unnecessary navigation and inefficient utilization of computing resources when they wish to experiment with a different algorithm.

Placeholder

New default model deployment#

To further improve user experience for model deployment, the default deployment option in the platform is now set to SDK deployment, instead of utilizing the exportable code. This enables users to take advantage of the SDK functionalities to build inference pipelines directly, rather than just running live inference on their device with the exportable code.

Key changes:

  • On the Models screen, only the raw models will now be exportable.

  • The Deployment screen remains the same, with the deployment package containing both the demo.py script that uses the SDK and the option for exportable code.

  • The different architecture options (‘fast’, ‘accuracy’, and ‘balanced’) will now be unified with the Models screen.

Please note, the existing ‘exportable code’ option for running inference will still be included in the SDK deployment package for advanced users who wish to deploy the model without using the Intel® Geti™ SDK.

Support for OpenVINO 2023.0#

The Intel® Geti™ platform supports OpenVINO 2023.0, the latest version of OpenVINO to date. You can export optimized models for deployment with OpenVINO 2023.0 to run on the latest Intel hardware.

Learn more about OpenVINO 2023.0 here.

System Requirements for Intel® Geti™ 1.8.0#

The platform can be installed on a machine with the following minimum hardware requirements:

  • CPU for workstations: Intel® Core™ i7, Intel® Core™ i9 or Intel® Xeon® scalable processors family capable of running 20 concurrent threads (in case of using the default K3s) or 48 concurrent threads (in case of using pre-installed K8s).

Note

From Intel® Core™ family, we recommend the following CPU series:

  • 13th gen (Raptor Lake): Intel® Core™ i7 13700 series and Intel® Core™ i9 13900 series

  • 12th gen (Alder Lake): Intel® Core™ i9 12900 series

  • CPU for cloud deployments: CPUs capable of running min. 24 concurrent threads for K3s or min. 48 concurrent threads for K8s (so for example, on AWS EC2 instances, this requirement would be translated to min. 24 vCPUs for K3s or min. 48 vCPUs for K8s).

  • GPU: min. one NVIDIA GPU with min. 16GB of memory (e.g. RTX 3090, RTX 4080, RTX 6000, RTX 8000, Tesla A100, Tesla V100, Tesla P100, or Tesla T4), other NVIDIA GPUs in a similar series are likely also compatible if they meet minimum memory requirements. However, the full range of devices is not fully tested and not specifically supported. We recommend 24GB of memory for the stable training & optimization.

  • Memory: min. 64 GB RAM (128 GB recommended) per GPU

  • Disk Space: min. 500 GB (1 TB recommended) for a root partition

Installation of the platform on a multinode configuration of Kubernetes cluster is not supported.