This is a project that uses OPTICS clustering algorithm to cluster footballer faces and diabetic patients' data.
Install scipy==0.21dev0 with :
pip install git+https://github.com/scikit-learn/scikit-learn.git
unless OPTICS is now part of the stable releas
Install all the rest of the requirements with:
pip install -r requirements.txt
Datasets
*dataset_diabetes*
- diabetic_data.csv : 130 US hospital data from 1999 to 2008[1](https://archive.ics.uci.edu/ml/datasets/Diabetes+130-US+hospitals+for+years+1999-2008)
- IDs_mapping.csv : mapping for admission type, discharge disposition and admission source
*footballers* : 124 footballers' photo (Neymar Jr., Lionel Messi, Cristiano Ronaldo, Luis Suarez, and Mohamed Salah)
*Predict* : 5 footballers' photo to use for prediction (Neymar Jr., Lionel Messi, Cristiano Ronaldo, Luis Suarez, and Mohamed Salah)
Ressources
- footballers_encodings.pickle : Encodings of the footballers after using *encode_faces.py*
- footballers_predict_encodings.pickle : Encodings of the footballers for prediction after using *encode_faces.py*
- GUFD_encodings.pickle : Encodings of the GUFD photo after using *encode_faces.py*
- shape_predictor_68_face_landmarks.dat : Used to extract facial features in *encode_features.py*
-
encode_faces.py : used to encode the face in an image as a 128-d vector
-
encode_features.py : used to encode the facial features of a face as a 7-d vector
-
faces_clustering.py : used to cluster the footballers' faces
-
diabetic_clustering.py : used to cluster the diabetic's data
-
similarity_clustering.py : used to cluster the GUFD dataset based on facial features similiarities
-
optics.py : contains the clustering and predicting methods
-
requirements.txt : file containing all the python packages requirements
- Victor MEUNIER - DMFinalProject - MrEliptik
This project is licensed under the MIT License - see the LICENSE.md file for details