tpcd / lifelongreid Goto Github PK

Hi, i'm very interested in your project on lifelong re-id, would you like to share a link of your CVPR paper ? thanks.

Thanks for your sharing! It is a nice work!
I follow your new job "A Memorizing and Generalizing Framework for Lifelong Person Re-Identification" from your new project webpage.
But I find that there is a problem with the code link of MEGE. I want to know, will you release new codes for MEGE? or in this repo?

The link of VIPeR is expired, could you please offer me a download link by BaiduDisk or sth else?

Hi Nan,
Thanks for sharing code. I managed to reproduce the results on all SEEN domains, but found, except for Market1501, there are great differences camparing with the results reported in the paper. So, how can I reproduce the results on CUHK, DUKE and MSMT17? Do I need to modify the dataloader or something?

Thanks for your code. I managed to reproduce the results for Market, Duke and CUHK03. However, I tried to add CUHK-SYSU and MSMT17 into argument --test_dataset, which could not give results mentioned in the paper. The test_mode argument seems to be not working in test_p_s.py, while test_continual_operation_neck.py only support market/duke/mix. Could you please share more details on how to test CUHK-SYSU and MSMT17 datasets?

Thank you for your sharing! In the paper, you give the final performance of Market, but I am curious about the performance change of Market during training, i.e., will the performance on Market decrease after the training on SYSU?