作为视频监控领域的核心技术,ReID 目前正向着无监督学习、抗遮挡、防换装、大规模预训练、训练样本生成等方向发展。
✔ 基于部分的伪标签提精,用于无监督人员重识别
Part-based Pseudo Label Refinement for Unsupervised Person Re-identification
https://arxiv.org/abs/2203.14675
https://github.com/yoonkicho/PPLR
✔ 特征空间的样本生成,用于孤立摄像机人员再识别
Camera-Conditioned Stable Feature Generation for Isolated Camera Supervised Person Re-IDentification
https://arxiv.org/abs/2203.15210
✔ 带有噪声标签的人员重新识别(Re-ID)的大规模预训练问题
Large-Scale Pre-training for Person Re-identification with Noisy Labels
https://arxiv.org/abs/2203.16533
https://github.com/DengpanFu/LUPerson-NL
✔ 将真实世界人物的衣服贴合到3D角色上,用于生成ReID的合成样本
Cloning Outfits from Real-World Images to 3D Characters for Generalizable Person Re-Identification
https://arxiv.org/abs/2204.02611
https://github.com/Yanan-Wang-cs/ClonedPerson
✔ 隐式样本扩展,用于解决无监督人员重识别中的聚类问题
Implicit Sample Extension for Unsupervised Person Re-Identification
https://arxiv.org/abs/2204.06892
https://github.com/PaddlePaddle/PaddleClas
✔ 基于图采样的深度度量学习用于可泛化的人员重识别
Graph Sampling Based Deep Metric Learning for Generalizable Person Re-Identification
https://arxiv.org/abs/2104.01546
https://github.com/ShengcaiLiao/QAConv
✔ 新Neighbor Transformer Network :克服身份内部变化,生成更具鉴别和鲁棒的特征
NFormer: Robust Person Re-identification with Neighbor Transformer
https://arxiv.org/abs/2204.09331
https://github.com/haochenheheda/NFormer
✔ 仅用RGB图像的换装人员再识别,克服衣服对识别的影响
Clothes-Changing Person Re-identification with RGB Modality Only
https://arxiv.org/abs/2204.06890
https://github.com/guxinqian/Simple-CCReID
✔ 用于遮挡人员重识别的特征擦除与扩散网络
Feature Erasing and Diffusion Network for Occluded Person Re-Identification
https://arxiv.org/abs/2112.08740
END