REMIND custom indoor dataset
收藏资源简介:
REMIND定制室内数据集是专为长期多目标重识别研究构建的基准数据集,由马德里理工大学计算机视觉团队创建,旨在模拟机器人室内导航中的对象再识别场景。该数据集包含精心设计的控制性重访序列,涵盖密集同类物体干扰环境,通过单目RGB视频捕捉静态物体在长时间间隔、显著视角变化及严重光照差异下的外观特征。数据采集过程模拟了机器人离开并重新进入房间的完整流程,构建了包含数百帧时间跨度的对象消失与重现序列。该数据集主要应用于计算机视觉与机器人领域,致力于解决通用室内物体在单目视觉下的长期身份一致性保持问题,为跨范式跟踪算法提供标准化评估基准。
The REMIND Custom Indoor Dataset is a benchmark dataset specifically constructed for long-term multi-object re-identification research. It was created by the Computer Vision Team of the Technical University of Madrid, aiming to simulate object re-identification scenarios in robotic indoor navigation. This dataset includes meticulously designed controlled revisit sequences covering environments with dense same-category object distractions, and captures the appearance characteristics of static objects under long time intervals, substantial viewpoint changes and severe lighting variations via monocular RGB videos. The data collection process simulates the complete workflow of a robot exiting and re-entering a room, constructing object disappearance and reappearance sequences with a time span of hundreds of frames. This dataset is primarily applied in the fields of computer vision and robotics, dedicated to solving the problem of long-term identity consistency maintenance of general indoor objects under monocular vision, and providing a standardized evaluation benchmark for cross-paradigm tracking algorithms.

- 1REMIND: RE-Identification with Memory for INDoor Navigation马德里理工大学·计算机视觉与航空机器人实验室; 西班牙高等科学研究委员会·自动化与机器人中心 · 2026年




