---
id: nuscenes-devkit
version: "1.2.0"
license: apache-2.0
license_treatment: noncommercial
maintenance: active
---

# nuscenes-devkit — The official devkit of the nuScenes dataset (www.nuscenes.org).

License: noncommercial · Maintenance: active · Downloads: 174.1K/mo

## What it is and what it does

The nuScenes devkit is the official toolkit for working with the nuScenes autonomous driving dataset. It provides Python APIs to load, visualize, and evaluate predictions against multi-sensor data (lidar, radar, camera) collected from autonomous vehicles. The package handles 3D object detection, multi-object tracking, trajectory prediction, and semantic/panoptic segmentation of point clouds.

The devkit depends on standard scientific Python libraries (numpy, scipy, scikit-learn, matplotlib, opencv-python-headless) plus specialized tools like pyquaternion for 3D rotations, Shapely for geometry, and pycocotools for evaluation metrics. It's designed to work with downloaded nuScenes dataset splits (train, val, test, mini) organized in a specific folder structure, and provides tutorials and evaluation code for benchmark challenges.

Use it for:

- Evaluate 3D object detection models on the nuScenes benchmark dataset
- Develop and test autonomous driving perception algorithms using standardized multi-sensor data
- Analyze lidar point cloud semantic segmentation with panoptic labels
- Implement trajectory prediction models using the nuScenes prediction challenge framework
- Visualize and debug sensor fusion pipelines with synchronized camera, lidar, and radar data
- Benchmark tracking algorithms against ground-truth annotations across 40,000+ keyframes

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides tools to load, parse, and analyze the nuScenes autonomous driving dataset, including 3D object detection, tracking, prediction, and semantic segmentation of lidar and camera data.

Yes, if you are working with the nuScenes dataset or benchmarking autonomous driving perception systems. The package is actively maintained, has low install friction, and is the official toolkit for the dataset. However, verify that the non-commercial license restriction aligns with your use case before committing to production work.

## Install

pip install nuscenes-devkit
uv add nuscenes-devkit
poetry add nuscenes-devkit

## Installing nuscenes-devkit

Before you install: Low friction installation with a pure-Python wheel. The package is actively maintained with a recent release (Aug 2025) and supports current Python versions (3.9, 3.12). Repository shows steady activity with 2789 stars.

License in practice: Licensed under Apache 2.0 but marked as non-commercial use only. Commercial use of this devkit requires separate licensing; verify terms before deploying in production systems.

Quickstart:

pip install nuscenes-devkit

from nuscenes.nuscenes import NuScenes
ns = NuScenes(version='v1.0-mini', dataroot='/data/sets/nuscenes')
ns.render_sample(ns.sample[0]['token'])

Requires the nuScenes dataset to be downloaded and extracted to a local directory (e.g., /data/sets/nuscenes); the devkit alone is not functional without dataset files.

Verify before relying:

- Whether the non-commercial license restriction applies only to the devkit code or also to derivative works built with it
- Dataset download and storage requirements beyond the devkit itself
- Performance characteristics when working with full-scale nuScenes data

## Package facts

- License: apache-2.0 (noncommercial)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 174.1K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

autonomous driving dataset toolkit, 3d object detection evaluation, lidar point cloud annotation, nuScenes data loader, self-driving car benchmark, panoptic segmentation lidar, trajectory prediction challenge, autonomous-driving, 3d-perception, dataset-toolkit

[View on SkillFed](https://skillfed.io/packages/nuscenes-devkit) · [View on PyPI](https://pypi.org/project/nuscenes-devkit/)