Skip to main content
Avala is Fleet Learning Infrastructure for Physical AGI: independent infrastructure between your real-world systems and your training stack. Every fleet should learn from its own experience. Annotation is one step of the loop that turns raw sensor recordings into versioned, traceable training artifacts. The customer trains and owns the model and retains authority over training and deployment. Annotation is one step of that loop. This section covers the labeling tools, project management, and team collaboration you use in Mission Control.

​
Features

​
Dataset Management

  • Upload: Drag-and-drop upload for images, videos, and point clouds
  • Organize: Group data by project, category, or custom tags
  • Browse: Grid and list views with filtering and search
  • Sequences: Manage video and temporal data sequences

​
Annotation Tools

Professional annotation tools for multiple data types:
  • 2D Annotation: Bounding boxes, polygons, polylines, segmentation
  • 3D Annotation: Cuboids in point cloud data with multi-camera projection
  • Video Annotation: Frame-by-frame labeling with object tracking
  • Multi-Camera: Synchronized annotation across camera views

​
Project Builder

Visual workflow builder for annotation projects:
  • No-code configuration: Drag-and-drop project setup
  • Label taxonomy: Define object classes and attributes
  • Quality stages: Configure review and approval workflows
  • Templates: Reuse project configurations

​
Quality Control

Built-in quality assurance features:
  • Review dashboard: Accept or reject annotations
  • Issue tracking: Flag and resolve problems
  • Metrics: Track annotation quality and consistency
  • Reports: Generate quality reports

​
Team Collaboration

  • Organizations: Group users and resources
  • Teams: Create focused workgroups
  • Permissions: Role-based access control
  • Activity tracking: Monitor project progress

​
Dataset Views

When viewing a dataset:

​
Get Started