A Python API for the Argoverse dataset, providing tools for 3D tracking, motion forecasting, and HD map interaction for autonomous vehicle research.
Argoverse API is a Python library that provides tools to access and manipulate the Argoverse dataset, a collection of sensor data, high-definition maps, and annotations for autonomous vehicle research. It enables loading of tracking and forecasting data, map queries, and visualization to support development of perception and prediction models.
Researchers and engineers working on autonomous vehicle perception, motion forecasting, and 3D tracking who need a structured way to access and experiment with large-scale driving datasets.
It offers a unified, well-documented interface to a rich multimodal dataset with HD maps, reducing preprocessing overhead and accelerating research iteration compared to handling raw data directly.
Official GitHub repository for Argoverse dataset
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Consolidates access to both Argoverse-Tracking and Argoverse-Forecasting datasets with consistent methods for loading sensor data, poses, and annotations, reducing preprocessing overhead.
Provides tools to interact with city-scale maps, enabling queries for lane geometry, traffic signals, and drivable areas, which are crucial for map-based perception tasks.
Includes scripts and Jupyter notebooks for rendering birds-eye-views, cuboids on images, and ground lidar points, aiding in data exploration and debugging.
Handles SE(3) and SE(2) transformations between city, egovehicle, and camera frames, simplifying complex spatial operations without custom calibration code.
The API is tightly coupled with the Argoverse dataset; adapting it for other datasets requires significant modification, as noted by the separate converters for nuScenes and Waymo.
Installation involves multiple steps: cloning the repo, downloading and extracting large HD map and dataset files separately, and managing optional dependencies like Mayavi and Open3D.
The README highlights the newer Argoverse 2 API, suggesting this version might be less maintained, potentially leading to compatibility issues with modern Python environments or datasets.