Pytorch DL Tutorial For Students

This repository contains a Jupyter notebook called TUTORIAL.ipynb that shows how to train a deep learning model in PyTorch to classify dog breeds. The notebook uses the ResNet-18 architecture and is trained on a subset of Stanford Dogs Dataset which contains 120 breeds of dogs.

Requirements

Python 3.6 or higher See requirements.txt for required libraries

Preparations

Creating a Conda Environment and Installing Libraries with Pip

These instructions will guide you through creating a new Conda environment called dltut, installing pip with Conda, and using pip to install libraries from a requirements file.

Prerequisites

• Conda is installed on your system. If not, you can download and install it from the Anaconda website.

Steps

1. Open anaconda prompt on your computer (Start button --> type Anaconda prompt). You should now see (base) at the start of your prompt.
2. Create a new Conda environment called "dltut" by typing the following command: conda create --name dltut
3. Activate the new environment by typing: conda activate dltut
4. Install pip with Conda by typing: conda install pip -y
5. Install some other jupyter notebook dependencies: conda install nb_conda_kernels ipywidgets ipykernel -y

Now you are ready to go the the Usage section.

Usage

1. Clone the repository: git clone https://gitlab.kuleuven.be/mebios-dl/masterpracticum_2023/pytorch-dl-tutorial-for-students.git
   (in case you don't have git installed, you can download it from here).
2. Install the required libraries by running the following command: pip install -r requirements.txt (you have to run this command from within the repository folder).
3. Extract dogs.zip into the project directory (if on Windows, otherwise the notebook will extract it for you).
4. Start Jupyter Lab and open the TUTORIAL.ipynb file: jupyter-lab (from the project folder)
5. Follow the instructions in the notebook to train and test the model.

Files

TUTORIAL.ipynb: Jupyter notebook with the tutorial code
dogs.zip: Compressed file containing the subset of the Stanford Dogs Dataset
requirements.txt: List of required libraries

Credits

The ResNet-18 implementation is adapted from the official PyTorch documentation: https://pytorch.org/docs/stable/_modules/torchvision/models/resnet.html The Stanford Dogs Dataset is provided by the Stanford Vision Lab: http://vision.stanford.edu/aditya86/ImageNetDogs/