first commit

app.py

@@ -0,0 +1,74 @@
+
+import gradio as gr
+import os
+import torch
+
+from demos.model import create_model
+from timeit import default_timer as timer
+
+
+# Setup Class Names
+class_names = [
+    "glioma",
+    "meningioma",
+    "notumor",
+    "pituitary"
+]
+
+# Create efficientnet model
+efficient_model, transforms = create_model(num_classes=len(class_names))
+
+# Load saved weights for MRI classifier
+state_dict = torch.load(
+    f="/content/demos/efficientnet_mri_model.pth",
+    weights_only=False,
+    map_location="cpu"
+)
+
+efficient_model.load_state_dict(state_dict())
+
+def predict_img(img):
+
+  # Start Timer
+  start_time = timer()
+
+  # Transform image and add extra dimension
+  img = transforms(img).unsqueeze(0)
+
+  # Put the model on eval and inference mode
+  efficient_model.eval()
+  with torch.inference_mode():
+    
+    # Pass the transformed image through the model.
+    pred_probs = torch.softmax(efficient_model(img), dim=1)
+
+    # Pred Labels and Probabilities required in gradio format
+    pred_labels_and_probs = {class_names[i] : float(pred_probs[0][i]) for i in range(len(class_names))}
+
+    # Calculate the total time
+    pred_time = round(timer() - start_time, 5)
+
+    return pred_labels_and_probs, pred_time
+
+# Ddfine the title, desc and article
+title = "Dr. Gerry MRI result classified"
+desc = "An Efficientnet model feature extractor to classify MRI images"
+article = "Created @ Mauaque Resettlement Center Gonzales Compound"
+
+# Create example list
+path_list = '/content/demos/examples'
+example_list = os.listdir(path_list)
+
+# Initialized Gradio Demo
+demo = gr.Interface(
+    fn=predict_img,
+    inputs=gr.Image(type="pil"),
+    outputs=[gr.Label(num_top_classes=4,label="Predictions"),
+             gr.Number(label="Prediction's time")],
+    examples=example_list,
+    title=title,
+    description=desc,
+    article=article
+)
+
+demo.launch()

efficientnet_mri_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0fbe8aef55563e04f1af41ee2e66ea3d13b7f3ad108823c73a35310392c2722b
+size 31365477

model.py

@@ -0,0 +1,22 @@
+
+import torchvision
+import torch
+
+def create_model(
+    num_classes: int=4,
+    seed: int=42):
+
+
+  weights = torchvision.models.EfficientNet_B2_Weights.DEFAULT
+  transform = weights.transforms()
+  model = torchvision.models.efficientnet_b2(weights=weights).to("cpu")
+
+  for params in model.parameters():
+    params.requires_grad = False
+
+  model.classifier = torch.nn.Sequential(
+      torch.nn.Dropout(p=0.2,inplace=True),
+      torch.nn.Linear(1408,num_classes)
+  )
+
+  return model, transform

requirements.txt

@@ -0,0 +1,3 @@
+torch==2.10.0
+torchvision==0.25.0
+gradio==5.50.0