add files

app.py

@@ -0,0 +1,54 @@
+import os
+os.environ['CUDA_VISIBLE_DEVICES'] = '-1'
+import tensorflow as tf
+import keras.backend.tensorflow_backend as tb
+tb._SYMBOLIC_SCOPE.value = True
+import numpy as np
+import gradio as gr
+import cv2
+from PIL import Image
+from tensorflow.keras.models import load_model
+
+# Get model weights
+#os.system("wget https://github.com/hasibzunair/adversarial-lesions/releases/latest/download/MelaNet.h5")
+
+# Load model
+model = None
+model = load_model("MelaNet.h5", compile=False)
+model.summary()
+
+examples = None #["./examples/covid.jpeg", "./examples/normal.jpeg"]
+labels = ["Benign", "Malignant"]
+
+# Helpers
+def preprocess_image(img_array):
+    # Normalize to [0,1]
+    img_array = img_array.astype('float32')
+    img_array /= 255
+    # Check that images are 2D arrays
+    if len(img_array.shape) > 2:
+        img_array = img_array[:, :, 0]
+    # Convert to 3-channel
+    img_array = np.stack((img_array, img_array, img_array), axis=-1)
+    # Convert to array
+    img_array = cv2.resize(img_array, (224, 224))
+    return img_array
+
+
+def inference(image):
+    img = preprocess_image(img)
+    preds = model.predict(img)
+    img = np.expand_dims(img, 0)
+    # Predict
+    preds = model.predict(image)[0]
+    labels_probs = {labels[i]: float(preds[i]) for i, _ in enumerate(labels)}
+    return labels_probs
+
+gr.Interface(
+    fn=inference,
+    title="Benign or Malignant",
+    description = "Predict if this image has benign or malignant symptoms",
+    inputs="image",
+    outputs="label",
+    examples=examples,
+).launch(debug=True, enable_queue=True)

requirements.txt

@@ -0,0 +1,8 @@
+scipy==1.4.1
+tensorflow==2.2.0
+keras==2.3.1
+h5py==2.10.0
+numpy==1.18.1
+opencv-python-headless==4.2.0.32
+Pillow
+