# fundamentalvision/BEVFormer

[ECCV 2022] This is the official implementation of BEVFormer, a camera-only framework for autonomous driving perception, e.g., 3D object detection and semantic map segmentation.

Repository: https://github.com/fundamentalvision/BEVFormer
Canonical: https://ross.abutalabs.com/products/bevformer
Homepage: https://arxiv.org/abs/2203.17270
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, autonomous-driving, computer-vision, object-detection
Last push: 2024-08-15T05:08:16+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1617, "days_push": 748, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4579, forks 750 (observed 2026-08-28T04:08:54.507579+00:00)

## What it is
BEVFormer is the official PyTorch implementation of an ECCV 2022 paper that learns bird's-eye-view (BEV) representations from multi-camera images using spatiotemporal transformers. It supports camera-only autonomous driving perception tasks such as 3D object detection and map segmentation, achieving state-of-the-art results on nuScenes.

## Use cases
- run 3d object detection from camera images on nuScenes
- learn bird's eye view representation from multi-camera images
- segment driving maps from surround-view cameras
- benchmark camera-only perception against lidar baselines
- build a baseline for autonomous driving perception research
- estimate object velocity from cameras without lidar

## When to choose
- you need a proven camera-only 3D detection baseline for autonomous driving research
- you want to reproduce or extend a well-cited BEV perception method on nuScenes
- your perception stack must work without lidar, using only surround cameras

## When to avoid
- you need a production-ready, actively maintained perception stack for a deployed vehicle
- you lack multi-GPU hardware, since training requires significant GPU memory
- you need lidar or radar fusion rather than camera-only perception

## Facets
- artifact type: library
- maturity: maintenance
- function: deep-learning, computer-vision, image-processing
- domain: autonomous-vehicles, computer-vision, deep-learning, machine-learning
- platform: python
- tags: 3d-object-detection, bird-eye-view, nuscenes, transformer, multi-camera-perception, semantic-segmentation, research-code, eccv-2022, gpu, linux

## Member repositories
- fundamentalvision/BEVFormer (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:54.507579+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:19:54.483539+00:00, confidence not recorded.
  - readme: https://github.com/fundamentalvision/BEVFormer (fetched 2026-08-28T04:08:54.507579+00:00, sha 151921c501e4)
  - homepage: https://arxiv.org/abs/2203.17270 (fetched 2026-08-29T09:05:24.624256+00:00, sha ced45b9d92b0)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T09:05:24.627615+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T09:05:24.632580+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T09:05:24.634723+00:00, sha a1f16f915a9a)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T09:05:24.629992+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.