Add several person retrieval papers.

[layumi]

Dear author,

Thank you for your great collection.
I add several my papers about person re-identification into your list, which is a subtask of image retrieval.
Hope they are good to be merged in your list. Thanks a lot.

[willard-yuan]

@layumi Thank you for your pull request. The project collects papers for general image retrieval. I'm not sure all of the papers you added are for general image retrieval. Rid has some special tricks and some of the tricks are not suitable for general image retrieval. So please make sure the papers you added are suitable for general image retrieval. Thanks.

[layumi]

Hi @willard-yuan
Thanks a lot for your quick response.
I see you also included some fashion datasets and landmark datasets (Of course, landmark is one basic application). So would you like to consider the person reid as an application for image retrieval? Some re-id paper also run the general image retrieval experiments.

• A Discriminatively Learned CNN Embedding for Re-identification,TOMM2017
  I evaluate the method on Oxford 5k and compare the conventional method including MAC and RMAC.

• Unlabeled samples generated by gan improve the person re-identification baseline in vitro, ICCV2017
  Besides the person reid dataset, I also evaluate the proposed method on CUBird dataset, which is a fine-grained dataset.

• Pedestrian Alignment Network for Large-scale Person Re-identification, TCSVT2018.
  In the sprit of the region-based image retrieval methods, I propose an auto-alignment and recognition method.

• Joint Discriminative and Generative Learning for Person Re-identification (Oral)[Project] [YouTube] [Bilibili] [Poster], CVPR2019
  This is a CVPR oral paper. It introduces the data augmentation to person re-id. I think it will also works on face retrieval and other applications with cropped regions.

I really appreciate your helps.