nv-tlabs/lift-splat-shoot

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1369 stars · 260 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

PyTorch implementation of Lift-Splat-Shoot (ECCV 2020), an end-to-end model that converts images from arbitrary multi-camera rigs into a bird's-eye-view representation for autonomous vehicle perception. It lifts per-camera image features into 3D frustums, splats them into a BEV grid, and supports tasks like BEV object/map segmentation and interpretable motion planning.

Use cases

• generate bird's-eye-view segmentation from multi-camera images
• fuse predictions from multiple vehicle cameras into one BEV grid
• train a BEV perception model on nuScenes
• run end-to-end motion planning from camera images
• compare camera-only BEV models against lidar-depth baselines
• research camera-to-BEV unprojection methods

When to choose

• you need a camera-only bird's-eye-view perception baseline for autonomous driving research
• you work with nuScenes multi-camera data and want a proven ECCV 2020 method
• you want interpretable end-to-end planning via BEV cost maps

When to avoid

• you need a production-ready, actively maintained perception stack
• you require lidar or radar fusion rather than camera-only input
• you need support for datasets other than nuScenes out of the box

Facets

library · maturity maintenance

machine-learning deep-learning computer-vision image-processing autonomous-vehicles computer-vision deep-learning machine-learning python bird's-eye-view pytorch autonomous-driving nuscenes multi-camera-fusion 3d-perception eccv-2020 research-code

Member repositories

For agents

markdown · JSON · MCP: product_card(name="nv-tlabs/lift-splat-shoot")