基于学习体型表征的行人再识别

行人再识别是计算机视觉领域的一项重要任务，但大多数现有模型很大程度上依赖于颜色外观。针对目前很少研究解决目标人物衣服不一致的行人再识别问题，提出一种新的表征学习模型。该模型通过对抗性学习和特征分离来产生不受服装颜色或图案影响的体型特征表示。同时，由于缺乏包含同一个人服装变化的行人再识别数据集，创建了一个合成数据集来模拟服装变化。4个数据集(两个基准行人再识别数据集，一个跨模态行人再识别数据集，合成数据集)的定量和定性结果证实了该方法对几种最先进的方法的鲁棒性和优越性。

Person Re-identification Based On Learning Shape Representations

Person re-identification is an important task in computer vision,but most existing models rely heavily on color appearance.Aiming at the current little research solve the problem of person re-identification with inconsistent clothes, a new representational learning model is proposed.It is used to generate a body shape feature representation without being affected by clothing color or patterns via adversarial learning and feature disentanglement. At the same time, due to the lack of large-scale re-ID datasets which contain clothing changes for the same person, a composite dataset is created to simulate clothing changes.The quantitative and qualitative results across four datasets(two benchmark re-ID datasets,a cross-modality re-ID datasets,composite dataset)confirm the robustness and superiority of the apporach against several state-of-the-art approaches.