融合感知算法测试数据集

为支撑云控融合感知算法性能测试，构建了涵盖多场景、多光照、多天气条件的融合
感知算法测试数据集。数据采集时间为 2024 年 1 月至 2025 年 10 月，包含图像数据集和点
云数据集。其中图像数据 164726 帧，点云数据 53148 帧，数据集总大小共 1 TB，数据格式
分别为 jpg 和 pcd。采集过程覆盖直道、弯道、匝道汇入汇出、拥堵路段等不同道路场景，
涵盖白天、傍晚等多样光照条件，以及晴天、阴天、雾天等不同天气状况。数据经设备校准、
人工筛选与校验等多轮质量控制，确保数据可靠性。该数据集可有效支撑云控融合感知算法
在多模态环境下的性能验证，为智能汽车云控平台相关技术研发和测试提供数据支撑。

To support the performance testing of cloud-controlled fusion perception algorithms, a test dataset for fusion perception algorithms covering diverse road scenarios, lighting conditions and weather conditions was constructed. The data was collected from January 2024 to October 2025, and includes image datasets and point cloud datasets. Specifically, there are 164,726 frames of image data and 53,148 frames of point cloud data. The total size of the dataset is 1 TB, with the data formats being JPG and PCD respectively. The data collection covers various road scenarios including straight roads, curves, ramp merging and diverging, and congested road sections, as well as diverse lighting conditions such as daytime and dusk, and different weather conditions like sunny, cloudy and foggy days. The data has undergone multiple rounds of quality control procedures including equipment calibration, manual screening and verification to ensure its reliability. This dataset can effectively support the performance verification of cloud-controlled fusion perception algorithms in multi-modal environments, and provide data support for the research, development and testing of technologies related to intelligent vehicle cloud-controlled platforms.

搜集汇总

背景与挑战

背景概述

该数据集旨在支持云控融合感知算法的性能测试，包含图像和点云数据，覆盖多种道路场景、光照条件和天气状况，并经过严格质量控制。它可为智能汽车云控平台的技术研发与测试提供可靠的多模态数据支撑。