This is the implementation of a convolutional Neural Network on COVIDx CT
-
The following dataset contains CT images from 194,922 CT slices from 3,745 patients and 201,103 CT slices from 4501 patients respectively.
-
Classes are zero-indexed with Normal=0, Pneumonia=1, and COVID-19=2
-
The COVIDx CT-2 dataset is released as a directory of images (2A_images) and associated label files ({train,val,test}_COVIDx_CT-2A.txt) indicating classes and bounding boxes for the body region.
Link to download the dataset: Here
- Train Loss: 0.026 | Accuracy: 99.146
- Test Loss: 1.190 | Accuracy: 76.404
- Train Loss: 0.024 | Accuracy: 99.229
- Test Loss: 1.056 | Accuracy: 79.012
- Train Loss: 13.014 | Accuracy: 70.982
- Test Loss: 12.807 | Accuracy: 58.678