UVT2000
收藏资源简介:
UVT2000数据集由安徽大学信息材料与智能感知安徽省重点实验室创建,包含2000对直接从各种真实世界场景中捕获的未对齐的可见光和热图像对。该数据集旨在支持无对齐RGBT显著对象检测的研究,通过提供未经过人工对齐的图像对,反映了实际应用中的对齐问题。创建过程中,图像对未经过任何对齐处理,直接从热红外和CCD摄像机捕获,确保了数据集的自然性和实用性。UVT2000数据集的应用领域包括智能视频监控、遥感图像分析和无人机监控等,旨在解决由于图像对未对齐导致的显著对象检测性能下降的问题。
The UVT2000 dataset was developed by the Anhui Provincial Key Laboratory of Information Materials and Intelligent Sensing, Anhui University. It contains 2000 pairs of unaligned visible and thermal infrared images directly captured from diverse real-world scenarios. This dataset is constructed to support research on unaligned RGBT salient object detection, as it provides unaligned image pairs that reflect the alignment issues prevalent in real-world applications. During its creation, the image pairs were not subjected to any alignment processing and were directly captured using thermal infrared and CCD cameras, ensuring the dataset's naturalness and practicality. The application domains of the UVT2000 dataset include intelligent video surveillance, remote sensing image analysis, unmanned aerial vehicle (UAV) surveillance, and more, with the goal of addressing the performance degradation of salient object detection caused by misaligned image pairs.




