Predicting Noval Corona, Pneumonia, Healthy from X-ray scans using Keras and Scikit Learn.
In this project, I am using X-ray images to analyze the health of the patient lungs since COVID-19 attacks the epithelial cells (cells that come from surfaces of the respiratory tract) and X-ray scans are easy to get than CT scans.
The COVID-19 virus spreads by droplets produced as a result of coughing or sneezing of a COVID-19 infected person. This could happen either by direct close contact with an infected person or by indirect contact like the droplets on surfaces and clothes.
Note: I have made Training_multiple_models.ipynb private to avoid misuse, contact me @v.snehith999@gmail.com for complete directory ☺
The trained model is deployed using Streamlit module on local system.
- Go ahead and download the training and testing dataset from the link given
[Drive Link]https://drive.google.com/drive/folders/1aA-RK0ImIAtKkGqg_-I0rAOsEOZRSSQZ?usp=sharing).
Note: This is an year old dataset that I have managed to collect. Rightnow there are n number of standerdized datasets avilable online. A quick google seach will do the work. - Detailed blog on working and implementation:
Medium Blog
| 📂 Dataset | 📑 Blog |
|---|---|
| Dataset | Medium |
| 📈 Model | 🛠 model_loss | 💡 model_acc |
|---|---|---|
| Custom Model |  |
 |
| RESNET50 |  |
 |
| VGG16 |  |
 |
- To this end, I am open sourcing my code, model, and results. You can access it here: Code.
- Out of three models(Custom model, VGG 16, ResNet50), Custom model and VGG16 are performing better.
- Wise picking of Augmentation features lead to robust dataset generation.
- Introducing Batch Normalization would uplift model performance.
- Due to time constraints, I am unable to implement all of my thoughts, but I am planning out to make an end-to-end Web based application hosted on heroku
- Updating datasets as per the avilability.
- Version-2 dataset is collected during the July month-end. I will keep updating dataset as more and more data gets open-sourced. NOTE: If anybody got better metrics than me, please do share your knowledge after all Knowledge is Sharing!!
This blog post on COVID-19 prediction is a proof-of-concept and for educational purposes only. It is not meant to replace professional medical advice.
Linkedin: https://www.linkedin.com/in/snehitvaddi/
Github: https://github.com/snehitvaddi
https://towardsdatascience.com/using-deep-learning-to-detect-ncov-19-from-x-ray-images-1a89701d1acd https://jovian.ai/v-snehith999/covid19-training-different-models-v2