Projects
Projects are the primary workspace for annotation. Each project combines a robot configuration, a label schema, and one or more datasets into a cohesive annotation environment.
What is a Project?
A project is a container within an organization that groups everything needed to annotate a set of robot trajectories:
- Datasets — The CSV files (and optional videos) containing trajectory data to be labeled.
- Label Schema — The vocabulary of labels (e.g., reach, grasp, lift) that can be applied to frames.
- Robot Configuration — The robot model used for 3D visualization (generic arm or URDF model).
Projects enforce consistency: all datasets within a project share the same label schema and robot configuration. This ensures uniform annotations across your entire dataset collection.
Creating a Project
To create a new project:
- Navigate to your organization dashboard and click New Project.
- Enter a project name (e.g., "Pick and Place Demos" or "Pouring Task Q1").
- Optionally add a description to help team members understand the project's purpose.
- Select the robot type for 3D visualization:
- Generic Arm — A configurable articulated arm where you specify link lengths and joint types.
- URDF Model — Select from built-in URDF models for accurate rendering of supported robots (KUKA iiwa14, Franka Panda, Kinova J2S6S200, UR10, UR5 + Gripper, dVRK Classic PSM1).
- Click Create Project to finalize.
Robot type can be changed later
You can reconfigure the robot type and model after creating the project. Go to the Robot Config tab to update the visualization settings at any time.
Project Tabs
Once inside a project, the interface is organized into tabs that provide access to different aspects of the project:
Datasets Tab
The default view when you open a project. Shows all uploaded datasets with their annotation progress. From here you can:
- Upload new datasets (CSV files and optional MP4 video).
- Open a dataset to start or continue annotating.
- View labeling progress (percentage of frames labeled).
- Delete datasets you no longer need.
Labels Tab
Define and manage the annotation vocabulary for this project. See Label Management for full details. Key actions include:
- Add new labels with name, color, and keyboard shortcut.
- Reorder labels to control display priority.
- Set export values for each label.
- Load from or save to label templates.
Robot Config Tab
Configure how your robot is visualized in the 3D viewer. See Robot Configuration for detailed documentation. This tab allows you to:
- Select the robot type (generic arm or URDF).
- Configure joint mappings from CSV columns to robot joints.
- Select from built-in URDF models with pre-loaded mesh assets.
- Preview the robot model in a 3D viewport.
Settings Tab
Manage project-level settings including name, description, and danger-zone actions like deletion.
Uploading Datasets
Datasets are uploaded from the Datasets tab within a project. Each dataset consists of:
- A CSV file (required) — Contains trajectory data with one row per frame and columns for joint angles, positions, or other measurements. See CSV Format for requirements.
- An MP4 video file (optional) — A recording synchronized with the CSV data for visual verification during annotation. See Video Upload for details.
To upload a dataset:
- Click Upload Dataset in the Datasets tab.
- Drag and drop your CSV file (or click to browse). Optionally add an MP4 video.
- Enter a name for the dataset (defaults to the filename).
- Click Upload. The file will be processed and frames will be extracted.
File size limits
The maximum file size per upload is 500 MB for both CSV and MP4 files across all plans. However, your plan's total storage quota still applies — Free accounts have 100 MB of total storage. See your organization's Analytics page for current usage and limits.
Project Settings
The Settings tab provides project management options:
- Rename — Change the project name and description at any time.
- Delete — Permanently remove the project and all its datasets, labels, and annotations. This action cannot be undone.
Deleting a project
Deleting a project permanently removes all datasets, annotations, and configuration within it. Exported CSV files you have previously downloaded are not affected, but the source data will be gone. Only organization Owners and Admins can delete projects.