Pretrained Model

[code10086web]

Hi,friends,I am wondering if you could provide me pretrained model.

[mznw]

Hello, you can download the weights in here.

Hi,friends,I am wondering if you could provide me pretrained model.

[code10086web]

Hello, you can download the weights in here.

Thank you very much! By the way, I am wondering if cuhk03 pretrained model is convenient to provide me.

[code10086web]

Hi,friends, Could you give me some tips? I obtained a lower result compared to your paper using pretrained model you provided.

My Environment

• future==0.18.2
• joblib==1.0.0
• mock==4.0.3
• numpy==1.20.0
• Pillow==8.1.0
• scikit-learn==0.24.1
• scipy==1.6.0
• six==1.15.0
• threadpoolctl==2.1.0
• torch==1.6.0
• torchvision==0.7.0

The following are my test results

Datasets	mAP	R1	mAP(paper)	R1(paper)
market1501(pretrained model provided by  the author)	87.92	95.3	88.2	95.7
dukemtmc(pretrained model provided by  the author)	78.78	89.4	79	89
dukemtmc(the model I trained )	76.64	88.28	79	89
cuhk03(detected)(the model I trained)	73.78	77.36	74.8	78.9
cuhk03(labeled)(to be trained)			76.9	78.9

In addition, src/datasets/cuhk03.py may have a small bug.

def camera(file_path):

    return int(file_path.split('/')[-1].split('_')[1][0])     the last 0 may be chang to 1  ([1][0]——>[1][1])

Thanks a lot. Wish you a Happy New Year!