LLCM
收藏资源简介:
该数据集名为LLCM,是最新推出的可视红外图像人物重识别(VI-ReID)基准,包含了30,921张来自713个独特身份的图片。测试集则包含了13,909张来自351个不同身份的图片。该数据集在评估中同时采用了可视至红外(VIS2IR)和红外至可视(IR2VIS)两种模式。此外,由于包含了更多从不同视角拍摄的图片、更为复杂的低光照条件以及更长的时间跨度,该数据集呈现出更大的挑战性。规模上,数据集共有30,921张图片,涉及713个身份;而测试集则包含13,909张图片,涉及351个身份。该数据集的任务是进行可见光与红外图像中的人物重识别。
The dataset named LLCM is a newly released benchmark for visible-infrared person re-identification (VI-ReID). It contains 30,921 images from 713 unique identities, and the test set consists of 13,909 images from 351 distinct identities. Two evaluation modes, visible-to-infrared (VIS2IR) and infrared-to-visible (IR2VIS), are adopted for assessment on this dataset. Furthermore, this dataset poses greater challenges due to the inclusion of more images captured from diverse perspectives, more complex low-light conditions, and a longer time span. In terms of scale, the full dataset comprises 30,921 images spanning 713 identities, while the test set includes 13,909 images covering 351 identities. The core task of this dataset is person re-identification in visible and infrared imagery.




