SNACK Image Clustering Project for ET4 Info Polytech Paris Saclay

Author: Marko Babic + Code base from the ML course at Polytech Paris Saclay

Step 1: Python Version

• Case 1: Have a Python version between 3.8 and 3.11 => install the requirements:

pip install -r requierements.txt

• Case 2: Use the 3.11 environment from the .venv folder in the code To do this, in a terminal from the project directory:

.venv\Scripts\activate

Then make sure the following commands are executed via the environment Example of the terminal once the env is used: (.venv) PS C:\Users... => Use pip if the libraries are not installed correctly:

pip install -r requierements.txt

Step 2: Run the Clustering Pipeline

• Place yourself in the src folder in the terminal
• Execute the command:

Step 3: Launch the Dashboard

• Place yourself in the src folder in the terminal
• Execute the command:
• If there is a problem when launching the dashboard with the Excel files => delete all files in the src/donnees folder then redo steps 2 and 3

Appendix: Regarding the Images

• Link to Snack photos: https://huggingface.co/datasets/Matthijs/snacks/tree/main => depending on the power of your PC, I advise you to use only the data in the validation folder. => in the src/constant.py folder, make sure that the "PATH_DATA" variable is the path to the folder containing the images to be clustered.