Acknowledgment

This is a project for the course: Deep Learning of the University of Trento.

Professors:

• Elisa Ricci
• Willi Menapace

Students:

• Giovanni Lorenzini 223715
• Simone Luchetta 223716
• Diego Planchenstainer 223728

Necessary packages

Using pip install the following packages.

$ pip3 install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio===0.9.0 -f https://download.pytorch.org/whl/torch_stable.html

$ pip3 install pandas

$ pip3 install tensorboardX

$ pip3 install numpy

Dataset

The dataset needs to be extracted in a folder called dataset, mantaining the internal structure unchanged.

Code execution

Before execution is necessary to set to true or false the variable is_training in the main.py. If setted to true the network will train, this takes some time. If setted to false, and providing a network model (network.pt) is possible to go directly to the test mode. It will compute the mAP and the two required files (classification_test.csv and reid_test.txt). Then it's possible to execute the code simply by calling python and the path of the main file.

$ python3 main.py