search

emrahbasaran / SPReID

Code for our CVPR 2018 paper - Human Semantic Parsing for Person Re-identification
MIT License
141 stars 32 forks source link

reproduce results for cuhk03 #10

Open af00731 opened 6 years ago

af00731 commented 6 years ago

Hi, I read your human semantic parsing paper for person re-id and trying to reproduce results for cuhk03. For time being I am evaluating results just for re-id without semantic parsing. I have some queries to confirm or ask from you:

1) have you used cuhk03 detected and labelled combined together? as I saw your txt files and it has images > 13164 or your have used any kind of offline augmentation?

2) have you evaluated results for 20 splits and then took the average?

3) for each test split, have you performed single shot single query evaluation taking only one sample in query and gallery?

4) have you changed anything or added any layer in resnet 50 and resnet 152?

5) Have you checked for any other distance metric like XQDA? As many of the papers have told that XQDA always gives better than euclidean.

looking forward to your response. thanks in advance,

af00731 commented 6 years ago

have you checked your any results for 224 by 224 images?

af00731 commented 6 years ago

referring to table 6, means combination with resnet152 (ft) given in table caption. then what does inception-V3(ft) mean? should'nt it be resnet152(ft)? also, l2 normalization + concatenation means concatenating fg and wofg variations?

emrahbasaran commented 6 years ago

Hi,

We merged the labeled and detected images for training. To evaluate the results, we followed the standart protocol of the dataset. We used the original settings for resnet models, you can find the details in the paper. For distance calculation and re-ranking, we did several experiments using different methods including XQDA. But, the results given in the paper are the best. In table 6, Inception-V3(ft)* means the combination of Inception-V3(ft) with resnet152(ft).

Best,

af00731 commented 6 years ago

Thanks for answering

af00731 commented 6 years ago

can you please mention that which setting have you used for evaluation? Single shot single query or Multishot single query?

ajwl-pmli commented 4 years ago

Hello, Do you know how to compute the rank1 and mAP? could you give me the code which compute the rank1 and mAP, My mail is pku1401210454@163.com, Thank you very much! @af00731