CPU-Oriented Multi-Cue Person Re-ID Soft-Biometric Fusion

This dataset contains person images captured by non-overlapping surveillance cameras for research on person re-identification. The dataset is designed for evaluating person re-identification methods in realistic multi-camera environments where individuals appear across different viewpoints, illumination conditions, poses, and background contexts, without camera overlap. It is especially intended for studying CPU-efficient person re-identification, soft-biometric fusion, and distributed model updating strategies under constrained computational settings. The image set includes samples of the same individuals observed from different non-overlapping cameras, enabling the analysis of cross-camera appearance variation and identity matching performance. This dataset may be used for benchmarking re-identification approaches, validating lightweight models, and supporting reproducible research in intelligent video surveillance and computer vision.本数据集包含由非重叠监控摄像头采集的人物图像，用于人物重识别（person re-identification）相关研究。 本数据集旨在针对无摄像头重叠的真实多摄像头场景下的人物重识别方法开展评估——在此类场景中，同一人物会呈现出不同视角、光照条件、姿态及背景环境差异。该数据集尤其适用于研究受限计算环境下支持CPU高效运行的人物重识别、软生物特征融合以及分布式模型更新策略。 该图像集包含来自不同非重叠摄像头的同一人物样本，可用于分析跨摄像头外观变化与身份匹配性能。本数据集可用于基准测试重识别方法、验证轻量级模型，并为智能视频监控与计算机视觉领域的可复现研究提供支撑。