Model is trained on https://www.kaggle.com/nikhilpandey360/chest-xray-masks-and-labels dataset.
We have preprocessed the data by running 'preprocess.py'. It will convert all images into specific dimensions.
'train/*' means all training images are placed in the 'train' folder.
For training the model,we have converted the dataset into .h5 file. To convert into .h5 file, I have uploaded createh5.py file.
Please change the name respective of .h5 file name in config.py
Model is trained by running the command - python3 train.py -opt momentum --name my_network
Pretrained Model - https://drive.google.com/file/d/1RDoVQmxNbJESOBkmXH6T_3T811Zq0K2s/view?usp=sharing
The model is trained on 60 epochs with diagnostic-step size as 30.
For Compressing and Decompressing the image we will run the command - python3 compress.py -rl -i input_image_path -o random_output_image_name
For deployment of the model -
After training the model download the checkpoint and save it in the directory where app.py is stored.
Run command - python app.py
Go to browser and type http://localhost:5000/ to get the view of the system.
For uploading image you should upload the preproccesed image.
Output image will be stored in the 'output' folder.
Hope you will enjoy it!!!