This is the Pytorch code of the paper ''Person Re-Identification Baseline Based on Attention Block Neural Architecture Search''.

• Install pytorch with python 3.7, pytorch==1.4.0, torchvision==0.5.0, CUDA==10.1.
• Clone this repository:

git clone https://github.com/Nicholasxin/Attention-NAS-ReID
  

• Before run or debug the code, the folder named 'data' should be built and put the ReID public database in it directly. Then, RUN or DEBUG the code.

• Firstly, run the script train_search.py. The parameters of the four searched attention blocks would be print in the terminal.
• Secondly, put the obtained 4 parameters in the script '.NAS/genotypes.py' manually.
• Finally, run the script train.py to implement the ReID task by the searched network architecture.

• The test process will run at the same time with training while running the script train.py.

• 31/05/2022: code released