initial commit

app.py

@@ -0,0 +1,54 @@
+import gradio as gr
+import os
+import torch
+
+from model import create_vit
+from timeit import default_timer as timer
+from typing import Tuple, Dict
+
+class_names = ["NORMAL", "PNEUMONIA"]
+
+vit_model, vit_transforms = create_vit(seed=42)
+
+vit_model.load_state_dict(
+    torch.load(
+        f="finetuned_vit_b_16_pneumonia_feature_extractor.pth", 
+        map_location=torch.device("cpu")
+    )
+)
+
+def predict(img):
+    start_timer = timer()
+    
+    img = vit_transforms(img).unsqueeze(0)
+    
+    vit_model.eval()
+    with torch.inference_mode():
+        pred_prob_int = torch.sigmoid(vit_model(img)).round().int().squeeze()
+        
+    if pred_prob_int.item() == 1:
+        class_name = class_names[1]
+    else:
+        class_name = class_names[0]
+            
+    pred_time = round(timer() - start_timer, 5)
+    
+    return class_name, pred_time
+
+title = "Detect Pneumonia from chest X-Ray"
+description = "A ViT feature extractor Computer Vision model to detect Pneumonia from X-Ray Images."
+article = "Access project repository at [GitHub](https://github.com/Ammar2k/pneumonia_detection)"
+
+example_list = [["examples/" + example] for example in os.listdir("examples")]
+
+demo = gr.Interface(fn=predict, 
+                    inputs=gr.Image(type="pil"),
+                    outputs=[gr.Label(num_top_classes=6, label="Predictions"), 
+                    gr.Number(label="Prediction time(s)")],
+                    examples=example_list,
+                    title=title,
+                    description=description,
+                    article=article
+                   )
+
+demo.launch()

finetuned_vit_b_16_pneumonia_feature_extractor.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1737cbcc3556394cf4f82fa7f28cb985d11b32212adeff79fad3d7259674b17a
+size 343270545

model.py

@@ -0,0 +1,17 @@
+import torch
+import torchvision
+
+def create_vit(seed: int=42):
+    weights = torchvision.models.ViT_B_16_Weights.DEFAULT
+    
+    transforms = weights.transforms()
+    
+    model = torchvision.models.vit_b_16(weights=weights)
+    
+    for param in model.parameters():
+        param.requires_grad = False
+        
+    torch.manual_seed(seed)
+    model.heads = torch.nn.Sequential(torch.nn.LayerNorm(normalized_shape=768),
+                                      torch.nn.Linear(in_features=768, out_features=1))
+    return model, transforms