Occluded-DukeMTMC

Occluded-DukeMTMC 包含 15,618 个训练图像、17,661 个图库图像和 2,210 个被遮挡查询图像。 Occluded-DukeMTMC 上的实验结果将证明我们的方法在 Occluded Person Re-ID 问题上的优越性，更不用说我们的方法不需要任何手动裁剪过程作为预处理。

The Occluded-DukeMTMC dataset contains 15,618 training images, 17,661 gallery images, and 2,210 occluded query images. Experimental results on Occluded-DukeMTMC will verify the superiority of our method for the Occluded Person Re-ID task, not to mention that our method does not require any manual cropping procedure as a preprocessing step.

搜集汇总

背景与挑战

背景概述

Occluded-DukeMTMC是一个专门针对遮挡行人重识别（Occluded Person Re-Identification）的数据集，包含超过35,000张图像，具体分为15,618个训练图像、17,661个图库图像和2,210个被遮挡查询图像。该数据集旨在评估和提升在遮挡场景下的行人重识别性能，其设计无需手动裁剪预处理，由悉尼科技大学和百度于2019年发布，相关研究论文和代码库已公开。