Update app.py

app.py

@@ -1,41 +1,43 @@
-
-import gradio as gr
 from transformers import pipeline
+import gradio as gr
+
+
+modelName = "Melanoma-Cancer-Image-Classification"
+hfUser = "Hemg"
+
+
+def prediction_function(inputFile):
+    # get user name of their hugging face
+    modelPath = hfUser + "/" + modelName
+    # takes some time
+    classifier = pipeline("image-classification", model=modelPath)
+
+    try:
+        result = classifier(inputFile)
+        predictions = dict()
+        labels = []
+        for eachLabel in result:
+            predictions[eachLabel["label"]] = eachLabel["score"]
+            labels.append(eachLabel["label"])
+        result = predictions
+
+        # Check if the image is out of context
+        if "out of context image" in result:
+            raise ValueError("Out of context image provided")
+    except Exception as e:
+        result = "no data provided!!"
+
+    return result
+
+
+# change modelName parameter
+def create_demo():
+    demo = gr.Interface(
+        fn=prediction_function,
+        inputs=gr.Image(type="pil"),
+        outputs=gr.Label(num_top_classes=2),
+    )
+    demo.launch(debug=True)
 
-# Load the image classification pipeline
-try:
-    pipeline = pipeline(task="image-classification", model="Hemg/Melanoma-Cancer-Image-Classification")
-except ValueError:
-    pipeline = None
-
-def predict(input_img):
-    if pipeline is None:
-        result = {"error": "Model not available"}
-    else:
-        try:
-            # Use the pipeline to make predictions
-            predictions = pipeline(input_img)
-            # Process the predictions
-            result = {p["label"]: p["score"] for p in predictions}
-            # Check if the labels are "Benign" or  "Malignant
-            if not any(label in result for label in ["Benign", "Malignant"]):
-                result =  "This is out of context image"
-        except:
-            # If an exception occurs, return a default result
-            result = "no data provided!!"
-    
-    # Return the input image and the result
-    return input_img, result
-
-# Create a Gradio interface
-gradio_app = gr.Interface(
-    predict,
-    inputs=gr.Image(label="Upload Image", sources=['upload', 'webcam'], type="pil"),
-    outputs=[gr.Image(label="Processed Image"), gr.Label(label="Result", num_top_classes=2)],
-    title="Benign or Malignant?"
-)
-
-# Launch the Gradio app
-if __name__ == "__main__":
-    gradio_app.launch()
 
+create_demo()