Upload a.py

a.py

@@ -0,0 +1,75 @@
+
+
+import io
+import numpy as np
+from PIL import Image
+import streamlit as st
+import tensorflow as tf
+import matplotlib.pyplot as plt
+from tensorflow import keras
+
+
+st.title('Brain \U0001F9E0 Tumor Detector ')
+
+st.subheader('Find out whether there is a tumor \U0001F534 (Glioma Meningioma Pituarie) in the brain (or) \
+         not \U0001F7E2')
+
+
+
+
+inp_t = st.file_uploader(label='Upload MRI here', accept_multiple_files=True)
+
+
+
+def load_img(path):
+        # reading file object and making it to pil image and to np array
+        img_l = []
+        for i in path:
+                img_byte = i.read()
+                img = Image.open(io.BytesIO(img_byte))
+                img = img.resize((64, 64), Image.ANTIALIAS)
+                if img.mode != 'L':
+                        img = img.convert('L')
+                img_arr = np.array(img, dtype='float32')/255
+                img_arr = np.expand_dims(img_arr, axis=-1)
+                img_l.append(img_arr)
+        img = np.stack(img_l)
+        return img
+
+# prediction
+
+
+def pred(img):
+    # Load TFLite model and allocate tensors.
+    model = keras.models.load_model('model2_weights.h5')
+    result = model.predict(img)
+
+    return result
+
+
+
+# if file is uploaded
+if inp_t:
+        img = load_img(inp_t)
+        result = ['Glioma', 'Meningioma', 'no tumor', 'Pituarie']
+
+        st.warning(
+            '** Uploaded {} images [View images in side Panel]'.format(img.shape[0]))
+
+        res = pred(img)
+        max_value = res[0][np.argmax(res)]*100
+        if (result[np.argmax(res)] == 'no tumor'):
+            st.subheader("\U0001F7E2 Model predicts there is {}  tumor with {:.2f} % confidence].\U0001F7E2".format(result[np.argmax(res)], max_value))
+        else:
+            st.subheader("\U0001F534  Model predicts there is {}  tumor with {:.2f} % confidence.\U0001F534 ".format(result[np.argmax(res)], max_value))
+
+
+        st.write('\n')
+
+        st.image(inp_t, width = 400)
+
+   
+        st.markdown('---')
+        st.error(
+            'Dont conclude by looking at predictions, just take them as a reference!!')
+