Guides

Learn the API progressively first, then choose a computer-vision outcome. The four API levels explain increasing control; capability guides show how those levels combine in complete workflows.

For lookup rather than a walkthrough, open the component catalog and choose an Application, Stage, operation, estimator, typed result or Geometry primitive by role.

Learn by API level

Applications

Run, reconfigure, inspect and persist a task wrapper.

Applications: solve the task first

Pipeline

Compose direct-sphere and projected-view stages.

Pipeline: compose the method

Spherical Ops

Use functions, immutable operations and Torch layers.

Spherical Ops: control the operation

Geometry

Control pixels, rays, views, interpolation and masks.

Geometry: control rays and projections

• The four API levels
  • Applications: solve the task first
  • Pipeline: compose the method
  • Spherical Ops: control the operation
  • Geometry: control rays and projections

Build by outcome

These guides start from a result and connect the relevant API levels, mathematical explanation, extensions and evidence.

• Process and project panoramic images
  • Start
  • Compose
  • Understand and extend
• Extract and match visual features
  • Start
  • Compose
  • Extend
  • Evaluate
• Estimate pose, depth and radial range
  • Pose and scale
  • Two-view dense range
  • Multiview surface reconstruction
  • Single-view range
  • Validate and consume
• Registration and panorama stitching
  • 1. Load real images and run the Application
  • 2. Preserve explicit support and validity
  • 3. Inspect and reconfigure the Pipeline
  • 4. Use calibration independently
  • 5. Use robust model fitting independently
  • Failure boundary
• Localize and associate semantic regions
  • Start
  • Build proposals
  • Associate views
  • Extend and evaluate
• Composed spherical API
  • 1. Applications
  • 2. Pipeline composition
  • 3. Operations and Torch
  • 4. Contracts, persistence and architecture
  • Remaining Experimental promotion blockers
• Detailed tutorials
  • Compose a new sphere-native method
  • Project and reconstruct panoramic views
  • Process directly on the sphere
  • Extract and match visual features
  • Estimate pose and radial range
  • Reconstruct and navigate
  • Localize and associate semantic regions
  • Understand strategies and evidence
• How-to guides
  • Choose modality, layout, and interpolation
  • Preprocessing Containers
  • Spherical image processing
  • Spherical features and matching
  • Analyze regions on a sphere
  • Evaluate spherical results reproducibly
  • Build and query a spatial-semantic graph
  • Estimate spherical pose and reconstruct multiple panoramas
  • Run spherical visual SLAM