Get Started#
Interact programmatically with the Intel® Geti™ platform to leverage its capabilities in handling datasets, annotations, and more.
Note
The base URL in the endpoints is your IP address or your DNS domain you provided during the installation.
Obtain access#
Before you can start interacting with the Intel® Geti™ API, you first must obtain access to the API by creating an API token.
To get the token:
Go the Intel® Geti™ user interface.
Go to the Account section on the side panel and click on Token tab.
Click Create. A dialog box will prompt you to specify an expiration date for the token.
Warning
After closing the dialog box, you will not be able to retrieve your token. So, remember to copy your token.
The token generated through this method is essential for authenticating requests to the Intel® Geti™ API. To utilize it, simply include this token in the request header of your subsequent API calls. This is done by adding x-api-key: <token> to the header of each request. This process ensures that your requests are securely authenticated.
Once you complete this, you are successfully authenticated and ready to interact with our REST API.
System Constraints and Limit Handling from the Intel® Geti™ Platform 2.0.0#
In order to secure our system and also to ensure reasonable performance for all users, we have defined the following constraints to the entities within the system. These limits apply to projects, labels, tasks, annotations, datasets, images and videos. With the release of the Intel Geti platform 2.0.0, any endpoint that can create or update these entities can now throw a ‘409’ (CONFLICT) response whenever the limits have been reached. Below, you will find a table with all limits.
Entity |
Constraint |
Value |
On-prem |
Cloud |
Description |
|---|---|---|---|---|---|
Number of projects per organization |
max |
1000 |
❌ |
✔️ |
Enforced at project creation and dataset import as new project |
Number of labels per project |
max |
500 |
✔️ |
✔️ |
Enforced at project creation, dataset import as new project and label management features |
Number of task in the task chain project |
max |
2 |
✔️ |
✔️ |
Enforced at project creation |
Number of datasets per project |
max |
1000 |
❌ |
✔️ |
Enforced at the dataset creation and dataset import as new dataset |
Number of media per project |
max |
100K (images and videos) |
❌ |
✔️ |
Enforced at media upload and dataset import |
Pixels per image |
max |
35,481,600 |
❌ |
✔️ |
Enforced at image upload and dataset import. |
Image size (pixels) |
min |
32 |
✔️ |
✔️ |
Enforced at image upload and dataset import |
Image size (pixels) |
max |
20000 |
✔️ |
✔️ |
Enforced at image upload and dataset import |
Video size (pixels) |
min |
32 |
✔️ |
✔️ |
Enforced at video upload |
Video width (pixels) |
max |
7860 |
✔️ |
✔️ |
Enforced at video upload |
Video height (pixels) |
max |
4320 |
✔️ |
✔️ |
Enforced at video upload |
Video duration (seconds) |
max |
10800 |
✔️ |
✔️ |
Enforced at video upload |
Number of annotation version per image/frame |
max |
100 |
❌ |
✔️ |
Enforced at annotation submission feature. |
Training dataset size |
max |
1000 |
❌ |
✔️ |
Enforced at training/optimize/hpo job and a warning will be shown at the affected jobs. |
If you want to learn more about the changes that occurred with the introduction of the Intel® Geti™ platform 2.0.0, read the release notes.
REST API Reference#
The REST API is written in OpenAPI specification (OAS). The OAS allows for a high level of flexibility regarding tooling and plugging into your current workflow. We provide you with the generated REST API reference, however, you can download the specification and use your preferred tool. All you need to do is download the OpenAPI specification (you can download a JSON file from here and import it to your tool of choice.