Upload 8 files

app.py

@@ -0,0 +1,57 @@
+import streamlit as st
+import tensorflow as tf
+from tensorflow.keras.layers import Input
+from tensorflow.keras.models import Model
+from tensorflow.keras.preprocessing.image import img_to_array
+import numpy as np
+from PIL import Image
+
+# Load the TensorFlow SavedModel as a layer
+bit_model = tf.keras.layers.TFSMLayer('flower_BiT_model', call_endpoint='serving_default')
+
+# Create a new Keras model
+inputs = Input(shape=(224, 224, 3))
+outputs = bit_model(inputs)
+model = Model(inputs=inputs, outputs=outputs)
+
+# Class labels (make sure they're in the same order as during training)
+class_labels = ['dandelion','daisy', 'tulip', 'sunflower', 'rose']
+
+def preprocess_image(img):
+    img = img.resize((224, 224))
+    img_array = img_to_array(img)
+    img_array = np.expand_dims(img_array, axis=0)
+    img_array /= 255.0
+    return img_array
+
+def predict_image(image):
+    img_array = preprocess_image(image)
+    prediction = model(img_array, training=False)
+    if isinstance(prediction, dict):
+        prediction = list(prediction.values())[0]
+    predicted_class_index = np.argmax(prediction[0].numpy())  # Assuming the output is a probability distribution or logits
+    predicted_class = class_labels[predicted_class_index]
+    return predicted_class
+
+# Streamlit UI
+st.title("5 Flower Classification 💐")
+st.write("This model is trained using Big Transfer (BiT) with 98% accuracy.")
+
+images = ["daisy.jpg", "dandelion1.jpg", "rose.jpg", "sunflower.jpeg", "dandelion2.jpg", "tulip.jpg"]
+captions = ["daisy", "dandelion", "rose", "sunflower", "dandelion", "tulip"]
+current_row = 0
+for _ in range(2):
+  cols = st.columns(3)
+  for col, image, caption in zip(cols, images[current_row:current_row+3], captions[current_row:current_row+3]):
+      col.image(image, caption=caption)
+  current_row += 3
+
+st.write('Upload a flower image from the web or use these sample images above. Lets see if it is a daisy, dandelion, rose, sunflower, or tulip!')
+
+uploaded_file = st.file_uploader("Choose an image...", type="jpg")
+
+if uploaded_file is not None:
+    image = Image.open(uploaded_file)
+    st.image(image, caption='Uploaded Image.', use_column_width=True)
+    label = predict_image(image)
+    st.header(f"This is ***{label}***!")

requirements.txt

@@ -0,0 +1,4 @@
+streamlit
+tensorflow
+numpy
+Pillow