Learning convolutional multi-level transformers for image-based person re-identification

#########################################################

This is repository contains the code for the paper:

Learning convolutional multi-level transformers for image-based person re-identification

(https://doi.org/10.1007/s44267-023-00025-8)

#########################################################

Installation

pip install -r requirements.txt

(we use /torch 1.10.1 /torchvision 0.11.2 /timm 0.4.9 for training and evaluation.）

Prepare Datasets

For using the codes, please download the public Market-1501 and Duke datasets.

Prepare ResNet-50 Pre-trained Models

Please download the ImageNet pretrained resnet50 model.

Training and Test

Run the train.py and test.py with your GPUs. You can get the same results in the above paper.

Citation

if you use this code for your research, please cite our paper:

Yan, P., Liu, X., Zhang, P. et al. Learning convolutional multi-level transformers for image-based person re-identification. Vis. Intell. 1, 24 (2023). https://doi.org/10.1007/s44267-023-00025-8

Contact

If you have any problems. Please concat

zhpp@dlut.edu.cn or yanpl@mail.dlut.edu.cn