September 30, 2025. Leads & Copy — Aeva (Nasdaq: AEVA) has launched AevaScenes, a new open dataset featuring synchronized, multi-sensor FMCW 4D LiDAR and camera data with object velocity measurements, semantic segmentation, tracking, and lane line annotations. The dataset aims to boost autonomous vehicle perception research and expand FMCW LiDAR adoption.
AevaScenes is the industry’s first open dataset of its kind, supporting innovation in object detection, semantic segmentation, motion forecasting, scene flow, and trajectory estimation. It is now available for academic and non-commercial use at scenes.aeva.com.
Key features include high-fidelity FMCW LiDAR data, rich multimodal sensor fusion, flexible field-of-view options, and ultra-long-range annotations up to 400 meters.
According to James Reuther, Chief Engineer at Aeva, AevaScenes will allow researchers to push the boundaries of autonomous driving by advancing detection and tracking and unlocking new approaches to motion understanding.
The dataset includes 100 curated sequences from the San Francisco Bay Area, with urban and highway driving data across day and night conditions; 10,000 frames of time-synchronized FMCW LiDAR and RGB camera data at 10Hz; a sensor suite of 6 Aeva FMCW LiDAR sensors and 6 high-resolution RGB cameras; 4K resolution camera specs; PCD point clouds, JPEG images, and JSON annotations; a total size of approximately 200 GB; and data captured using Aeva’s Mercedes Metris test vehicles.
The dataset is 50% highway / 50% urban, 50% day / 50% night, with all sequences captured in clear weather with dry road surfaces.
AevaScenes is publicly available for free today for non-commercial use at scenes.aeva.com.
Source: Aeva
