Upload 5 files

app.py

@@ -0,0 +1,37 @@
+import streamlit as st
+from PIL import Image
+
+from utils import Predict
+
+# Page settings
+st.set_page_config(
+    page_title="Image Classification",
+    layout="wide",
+    initial_sidebar_state="expanded"
+ )
+
+# Title
+st.title('Classification of Chest X-ray Images')
+
+# Upload file
+uploaded_file = st.file_uploader(label="Choose a file", type=['jpg', 'jpeg','png'])
+
+sidebar = st.sidebar
+
+if uploaded_file is not None:
+    image = Image.open(uploaded_file).convert('RGB')
+    # image = np.array(image)
+    predict = Predict()
+
+    col1, col2 = st.columns([0.5, 0.5])
+
+    #Col 1
+    with col1:
+        st.markdown('<p style="text-align: center;">Input Image</p>', unsafe_allow_html=True)
+        st.image(image, width=425, caption="X-Ray Image")
+
+    #Col 2
+    with col2:
+        pred,topk = predict.predict(image)
+        st.text("Prediction: "+ pred)
+        st.text("Confidence: "+ str(topk))

history.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8617c399a6ff43caf7dd8f1e4a0d44671608ce5bb7caab017073495a20fa894
+size 1007

model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec5f2a733488ccbdac6a9661278b89e927260720b732baf7136fd7fb32abae46
+size 96471677

requirements.txt

@@ -0,0 +1,3 @@
+streamlit
+torch
+torchvision

utils.py

@@ -0,0 +1,29 @@
+import torch
+from torchvision import transforms
+
+class Predict():
+    def __init__(self):
+        self.transform = transforms.Compose([
+            transforms.Resize(size=256),
+            transforms.CenterCrop(size=224),
+            transforms.ToTensor(),
+            transforms.Normalize([0.485, 0.456, 0.406],[0.229, 0.224, 0.225])
+            ])
+
+        self.model = torch.load("model.pt",map_location=torch.device('cpu'))
+        self.class_dict = {0: 'COVID', 1: 'Normal', 2: 'Pneumonia'}
+
+    def predict(self, image):
+        test_image_tensor = self.transform(image)
+        if torch.cuda.is_available():
+            test_image_tensor = test_image_tensor.view(1, 3, 224, 224).cuda()
+        else:
+            test_image_tensor = test_image_tensor.view(1, 3, 224, 224)
+        with torch.no_grad():
+            self.model.eval()
+            # Model outputs log probabilities
+            out = self.model(test_image_tensor)
+            ps = torch.exp(out)
+            topk, topclass = ps.topk(1, dim=1)
+        
+        return self.class_dict[topclass.cpu().numpy()[0][0]], topk.numpy()[0][0]