Uploaded December 2019 | Updated September 2026, 2 weeks ago
This is the final result of deployed in production Keras model with Flask in Google Cloud Console (GCC). The model analyzing a given X-ray of human lungs and return the prediction that a diagnosis could be Normal, Bacteria or Virus.
Important: with this video I do not explain how to construct convolutional neural networks (CNN), python programming skills, and not cover practical programming on artificial intelligent in details. On this video I introduce my roadmap for my portfolio project, how it being transformed from simple model in Jupyter Notebook to user friendly on even mobile phones web application with the help of Flask framework.
The accuracy of the model is approximately 80% (with the help of some regularization techniques as Dropout, L2 Regularization and Rmsprop with Adam, more about CNN optimizers: youtube.com/watch?v=YacPECoI5SY). The whole architecture of Convolutional Neural Network is provided in the first part of this video (in Jupyter Notebook).
For deploying I decided to use Flask instead of Django because of it simplicity and compatibility with Amazon Web Services (AWS) and Google Cloud Storage (GCS). That works fine. App Engine in GCS is a perfect tool for deploying a Keras model as web application. Do not forget to carefully upload your deployment environment and crate Python YAML file that the server understand what is configuration of your application.
Important part is to understand how to upload an image file to the server (Google bucket) with Flask. By this, I modified permission for app. bucket to let user write and read (when app makes a prediction) files there.
Some additional work with HTML, CSS and Javascript and the model front-end looks bright, simply and user friendly, even compatible with mobile phones.
Link to web application: flaskdeploy-xray.appspot.com
Github link to the project: github.com/vb100/Pneumonia-X-Rays-of-Human-Lungs-AI-project
Content of the video:
# Step 1. Build A.I. model (basen on convolutional neural network) in Jupyter Notebook. 0:43
# Step 2. Build Flask application in local environment and deploy it locally. 9:19
# Step 3. Deploy Flask application on Google Cloud Storage (by using Shell Editor). 19:41
# Step 4. Test your application. 25:57
# Step 5. Publish link of your web application to the world and try on mobile device. 28:15
I think that similar technology can be applied on Corona (COVID-19) domain too.
Wishes from Lithuania!
Vytautas.
#aiproject
#deeplearningproject
#xrayproject
This is the final result of deployed in production Keras model with Flask in Google Cloud Console (GCC). The model analyzing a given X-ray of human lungs and return the prediction that a diagnosis could be Normal, Bacteria or Virus.
Important: with this video I do not explain how to construct convolutional neural networks (CNN), python programming skills, and not cover practical programming on artificial intelligent in details. On this video I introduce my roadmap for my portfolio project, how it being transformed from simple model in Jupyter Notebook to user friendly on even mobile phones web application with the help of Flask framework.
The accuracy of the model is approximately 80% (with the help of some regularization techniques as Dropout, L2 Regularization and Rmsprop with Adam, more about CNN optimizers: youtube.com/watch?v=YacPECoI5SY). The whole architecture of Convolutional Neural Network is provided in the first part of this video (in Jupyter Notebook).
For deploying I decided to use Flask instead of Django because of it simplicity and compatibility with Amazon Web Services (AWS) and Google Cloud Storage (GCS). That works fine. App Engine in GCS is a perfect tool for deploying a Keras model as web application. Do not forget to carefully upload your deployment environment and crate Python YAML file that the server understand what is configuration of your application.
Important part is to understand how to upload an image file to the server (Google bucket) with Flask. By this, I modified permission for app. bucket to let user write and read (when app makes a prediction) files there.
Some additional work with HTML, CSS and Javascript and the model front-end looks bright, simply and user friendly, even compatible with mobile phones.
Link to web application: flaskdeploy-xray.appspot.com
Github link to the project: github.com/vb100/Pneumonia-X-Rays-of-Human-Lungs-AI-project
Content of the video:
# Step 1. Build A.I. model (basen on convolutional neural network) in Jupyter Notebook. 0:43
# Step 2. Build Flask application in local environment and deploy it locally. 9:19
# Step 3. Deploy Flask application on Google Cloud Storage (by using Shell Editor). 19:41
# Step 4. Test your application. 25:57
# Step 5. Publish link of your web application to the world and try on mobile device. 28:15
I think that similar technology can be applied on Corona (COVID-19) domain too.
Wishes from Lithuania!
Vytautas.
#aiproject
#deeplearningproject
#xrayproject
