Installation :

Tip : Make sure to install Numpy, Pandas, Matplotlib first and then proceed next.

Torch package
Tkinter
HeidiSQL
Grab a cup of coffee as these will take some time !
click here to start server in HeidiSQL and configure settings by setting username and password.

Get, set and go :

Download complete Project files using following command from git bash/ cmd (terminal):

git clone https://github.com/souravs17031999/Retinal_blindness_detection_Pytorch

[Note you need mainly these four things to get started :]

blindness.py model.py classifier.pt send_sms.py

Create a new database and table accordingly.
Then, Go to 'blindness.py' file and change some configuration settings according to your database.

connection = sk.connect(
    host="localhost",
    user="root",
    password="********",
    database="********"
)

Now, your DB server must be connected.
Finally, you also want 'classifier.pt' file which contains model's dictionary required when it is to be loaded.
Download here and put that file in the same directory and then modify the path accordingly in the 'model.py' file.

model = load_model('../Desktop/classifier.pt')

Finally, execute your 'blindness.py' file and your GUI must start (recommended to start this from your terminal and keep all your project files in same directory).
Upload the image and get your predictions.

If you want to get SMS on mobile for your predictions , then Create an account on Twilio by verifying your number.
Next, get your credentials from the Dashboard of twilio and use 'send_sms.py' to fetch API request and fill and replace your credentials. [Note, currently 'send_sms.py' is commented out, so uncomment everything before using it].
Then, uncomment following lines in 'blindness.py'
#from send_sms import * #send(value, classes)
Your messaging service should start and also you now see Message_Id printed on your terminal.

[Note : You can use sample images in the folder sampleimages which is taken from the original test dataset to test the system]