Projects define how you will create labels and manage them. When you create a project, you will first customize its specifications (i.e. annotation type, tags, object classes, etc.) then populate the project by adding data to it.
By adding data to a project, you will automatically create a label for each data file. A label is defined as a JSON file that stores all tags and relevant information about the data file. Simply put, Project + Data ⇒ Label!
Create a new project
1. Click +New Project on the top right corner of the Projects menu.
2. In the Create Project window that appears, do the following :
Enter a project name.
Enter a project description. (Optional)
Select a data type.
Image (New) — *You can try our new features such as Manual Validation and Consensus Labeling
Select annotation types.
Project Access (Optional)
Toggle on/off to make your project publicly available or private.
3. Image Category
You can create an image category and choose the response type (Multiple Selection, Multiple Choice, Free Response) from the first box on the left.
You can create category options on the second box.
You can create category group on the third box.
4. Object Detection
You can add an object class by clicking the + button (e.g. Car)
Class Property (Optional)
You can add properties to further define each object class (e.g. Car — Black, White, Red, etc.)
Class Group (Optional)
You can group object classes
Project creation is finished once you click Finish on the bottom left.
You can check the detailed information on each steps from below documentation.
A sample project is a project we provide pre-loaded with a large dataset so that users can understand how to use Suite’s features before loading their own data.
Each sample project includes a Custom Auto-Label AI. Creating a Custom Auto-Label AI for a project requires multiple steps, but you can try out the feature without creating one yourself through the sample projects.
How to Create
Click +New Project on the top right corner of the Projects menu, and select Sample Project on the dropdown menu.
Select one of the five sample projects.
Tiny Set-10: For the Suite Demo | 10 images, includes all annotation types
Retail: Detecting products on the display stand | 1000 images
Agriculture: MineApple Detecting apples | 1001 images
Industrial surface: Detecting defects in steel pipes | 774 images
3. Click Create.
All large dataset projects except for Tiny Set-10 takes about 10 minutes to finish creating. You can track the progress on the top notification window.
Duplicate a project
If you want to create multiple projects with the same configuration, you can use the project duplicate feature. This saves your time in creating new projects from the very beginning.
How to duplicate a project
1. In Project Overview, click the settings icon on the Project Configuration card.
2. Click Duplicate on the top right corner of the Project Configuration card.
3. Enter the new project name and click Ok to finish creating the project.
※ Note : The datasets linked to the previous project do not get duplicated to the new project. You will have to use the Data tab under Account view to assigned a particular dataset to the newly duplicated project.