This repo contains the inference code and the saved model for the covid19 detection using cxr images

• pytorch
• torchvision
• PIL
• numpy
• argparse
• opencv2 (if using opencv2 readable image like, .jpg, .jfif, .png, etc)
• pydicom (if using .dcm image)

1. Clone the repo
2. Open up the terminal and type the following:
   python3 inference.py --model MODEL_NAME_FOR_INFERENCING --image_path PATH_TO_TESTING_IMAGE
   
   

Note:
Saved Model weights are provided with this repo, in the saved_models folder so if using resnet18 no need to provide the --model parameter. Also if you don't have image to test you can still try out inferencing on the example image provided in the images folder. Also the Ground Truth value of the test image is given in the csv file label.csv in the images folder.

PATH_TO_TESTING_IMAGE: Must be with the full extension of the image i.e. with .jpg or .dcm or .jfif, etc.

MODEL_NAME_FOR_INFERENCING choices:

1. resnet18 -- accuracy achieved: 66%
   
2. LeNet -- accuracy acheived: 64.9%

NOTE: You can find the training code for generating these models here.