NTU-Outdoor-38
收藏资源简介:
NTU-Outdoor-38数据集是由南洋理工大学快速富对象搜索实验室创建的大规模户外人物再识别数据集,包含38个真实监控摄像头捕获的48,347张图像。该数据集特别关注摄像头性能不平衡问题,通过隐私感知的数据收集策略,确保了数据的合法性和可用性。数据集中的图像涵盖了视角、光照和分辨率的大范围变化,适用于研究摄像头间性能不平衡和提升人物再识别系统的鲁棒性。
The NTU-Outdoor-38 dataset is a large-scale outdoor person re-identification dataset developed by the Rapid and Rich Object Search Lab at Nanyang Technological University. It contains 48,347 images captured by 38 real-world surveillance cameras. This dataset specifically focuses on the issue of camera performance imbalance, and adopts privacy-aware data collection strategies to guarantee the legality and usability of the dataset. The images within the dataset cover extensive variations in viewpoint, illumination and resolution, making it applicable for researching inter-camera performance imbalance and enhancing the robustness of person re-identification systems.




