Create app.py

app.py

@@ -0,0 +1,47 @@
+import joblib
+
+dt_model = joblib.load('dt_model.joblib')
+rfc_model = joblib.load('rfc_model.joblib')
+abc_model = joblib.load('abc_model.joblib')
+knn_model = joblib.load('knn_model.joblib')
+
+from sklearn.metrics.cluster import normalized_mutual_info_score
+import cv2
+import numpy as np
+
+def preprocess_image(image):
+  grayscale_img = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
+  normalized_img = grayscale_img / 255.0
+  resized_img = cv2.resize(normalized_img, (28,28))
+  flattened_img = resized_img.flatten()
+  processed_img = np.array([flattened_img])
+
+  return processed_img
+
+def pred(input1, input2):
+  class_names = ["T-shirt/top", "Trouser", "Pullover", "Dress", "Coat", "Sandal", "Shirt", "Sneaker", "Bag", "Ankle boot"]
+  processed_image = preprocess_image(input1)
+
+  if input2 == "Decision Tree":
+    preds = dt_model.predict_proba(processed_image)[0]
+  elif input2 == "Random Forest":
+    preds = rfc_model.predict_proba(processed_image)[0]
+  elif input2 == "AdaBoost":
+    preds = abc_model.predict_proba(processed_image)[0]
+  elif input2 == "KNN":
+    preds = knn_model.predict_proba(processed_image)[0]
+
+  #print(class_names)
+  #print(preds)
+
+  return {label: float(pred) for label, pred in zip(class_names, preds)}
+
+
+input_module1 = gr.inputs.Image(label = "Input Sample Image")
+input_module2 = gr.inputs.Dropdown(choices=['Decision Tree', 'Random Forest', 'AdaBoost', 'KNN'], label = "ML Models Dropdown")
+
+output_module1 = gr.outputs.Label(label = "Predicted Class")
+
+gr.Interface(fn=pred, 
+             inputs=[input_module1,input_module2],
+             outputs=output_module1).launch(debug = True)