Skip to content

Latest commit

 

History

History
23 lines (20 loc) · 1.09 KB

README.md

File metadata and controls

23 lines (20 loc) · 1.09 KB

versioncnn1

pneumoniaCNN.ipynb : done with google Colaboratory Make sure to activate GPU (Modifier-> Paramètres du Notebook-> Accélérateur matériel-> GPU -> Enregistrer) Detects Pneumonia : val_acc': 0.6343749761581421, 'val_loss': 0.6910778284072876 test.py : done with IDLE contains some functions needed for the dataset ( converting images from dicom to jpg/ splitting..) Python version : 3.10 pydicom 2.3.0 Pillow 9.0.0 numpy 1.22.1 opencv-contrib-python 4.6.0.66 opencv-python 4.5.5.64 matplotlib 3.5.1 scipy 1.7.3

Dataset

Kaggle dataset: https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia?resource=download

all images should be resized the same to be able to work with CNN

images were transformed into size 64 * 64 (they were way larger but there wasn't enough memory )

Choosing batch size

start with a number and keep doubling it until the training time decreases ( if there is a runtime error , decrease the batch size)