Analyzing X-Ray Imaging of potential covid-patient using deep learning
The dataset was quite small (180 images each). I do not claim any diagonstic performance of the model this is simply an experiment inspired by this paper
- Tensorflow 2.0
- Keras
- Numpy
- CV2
pip install tensorflow
or pip install tensorflow-gpu
(For the GPU accelerated version)
pip install opencv-python
XRAY Images of COVID Patients vs Normal (PA View)
The datasets are from HERE and Kaggle
python3 test.py sample.jpeg models/100accuracy
The command above should print "Negative"
I was able to get a 100% accuracy on validation data, however it's likely due to a small dataset I did a 80/20 split for the dataset