Add a lot of changes

app.py

@@ -1,74 +1,24 @@
-import time
-from turtle import width
-import torch
-import folium
-import numpy as np
-import pandas as pd
 import streamlit as st
-from folium.plugins import MarkerCluster
-from streamlit_folium import folium_static 
-
+import streamlit as st
+from multiapp import MultiApp
+from apps import (
+    home,
+    main_model,
+    fine_tune
+)
 st.set_page_config(
         page_title="Ship Detection using YOLOv5 Medium Model",
         page_icon=":ship:",
         layout="wide"
     )
 
-st.write("# Welcome to Ship Detection Application! :satellite:")
-st.markdown(
-        """
-        This application is build based on YOLOv5 with extral large model. User just
-        upload an image, and press the 'Predict' button to make a prediction base on
-        a training model before.
-
-        ### For more information, please visit:
-
-        - Check out [my github](https://github.com/bills1912)
-        - Jump into YOLOv5 [documentation](https://docs.ultralytics.com/)
-
-    """
-    )
-
-st.write("## Ship Imagery Prediction")
-map_col1, map_col2, map_col3 = st.columns(3)
-
-ais = pd.read_csv("https://raw.githubusercontent.com/bills1912/marin-vessels-detection/main/data/MarineTraffic_VesselExport_2022-11-25.csv")
-ais_jakarta = ais[ais['Destination Port'] == 'JAKARTA']
-ais_list = ais_jakarta.values.tolist()
-f = folium.Figure(width=1000, height=500)
-jakarta_vessels = folium.Map(location=[-5.626954250925966, 106.70735731868719], zoom_start=8).add_to(f)
-ais_data = folium.FeatureGroup(name="marine_vessels")
-mCluster = MarkerCluster(name="Marine Vessels")
-for i in ais_list:
-  html = f"<h3>{i[1]}</h3> Vessel Type: {i[8]} </br> Destination Port: {i[2]} </br> Reported Destination: {i[4]} </br> Current Port: {i[6]}\
-          </br> Latitude: {i[10]} </br> Longitude: {i[11]}"
-  iframe = folium.IFrame(html)
-  popup = folium.Popup(iframe, min_width=250, max_width=300)
-  ais_data.add_child(mCluster.add_child(folium.Marker(location=[i[10], i[11]], popup=popup, icon=folium.Icon(color="black", icon="ship", prefix="fa"))))
-jakarta_vessels.add_child(ais_data)
-folium_static(jakarta_vessels, width=1370, height=700)
-
+apps = MultiApp()
 
-st.write("### Model evaluation:")
-eval_col1, eval_col2, eval_col3, eval_col4 = st.columns(spec=4)
-eval_col1.metric("Precision", "89.52%")
-eval_col2.metric("Recall", "83.54%")
-eval_col3.metric("mAP 0.5", "85.39%")
-eval_col4.metric("mAP 0.5:0.95", "62.63%")
+# Add all your application here
 
-uploaded_file = st.file_uploader("Choose a ship imagery")
-if uploaded_file is not None:
-    st.image(uploaded_file, caption='Image to predict')
-    folder_path = st.text_input("Image path",
-                                help="This field the image path field that the model will predict the object inside the image that we have uploaded",
-                                placeholder="Copy the path of image to this field")
+apps.add_app("Home", home.app)
+apps.add_app("YOLOv5x6 Model", main_model.app)
+apps.add_app("Fine-Tuning Model", fine_tune.app)
 
-prediction = st.button("Predict")
-if prediction:
-    ship_model = torch.hub.load('ultralytics/yolov5', 'custom', path="best.pt", force_reload=True)
-    results = ship_model(f"{folder_path}")
-    with st.spinner("Loading..."):
-        time.sleep(3.5)
-        st.success("Done!")
-    st.image(np.squeeze(results.render()))
-    results.print()
+# The main app
+apps.run()

apps/fine_tune.py

@@ -0,0 +1,31 @@
+import time
+import torch
+import numpy as np
+import streamlit as st
+
+def app():
+    st.write("## Fine-Tuning Model Prediction")
+    st.write("### Model evaluation:")
+    eval_col1, eval_col2, eval_col3, eval_col4, eval_col5 = st.columns(spec=5)
+    eval_col1.metric("Precision", "99.03%")
+    eval_col2.metric("Recall", "98.39%")
+    eval_col3.metric("F1-Score", "98.71%")
+    eval_col4.metric("mAP 0.5", "98.96%")
+    eval_col5.metric("mAP 0.5:0.95", "69.61%")
+
+    uploaded_file = st.file_uploader("Choose a ship imagery")
+    if uploaded_file is not None:
+        st.image(uploaded_file, caption='Image to predict')
+        folder_path = st.text_input("Image path",
+                                    help="This field the image path field that the model will predict the object inside the image that we have uploaded",
+                                    placeholder="Copy the path of image to this field")
+
+    prediction = st.button("Predict")
+    if prediction:
+        ship_model = torch.hub.load('ultralytics/yolov5', 'custom', path="apps/model/main_model.pt", force_reload=True)
+        results = ship_model(f"{folder_path}")
+        with st.spinner("Loading..."):
+            time.sleep(3.5)
+            st.success("Done!")
+        st.image(np.squeeze(results.render()))
+        results.print()

apps/home.py

@@ -0,0 +1,40 @@
+import time
+from turtle import width
+import torch
+import folium
+import numpy as np
+import pandas as pd
+import streamlit as st
+from folium.plugins import MarkerCluster
+from streamlit_folium import folium_static 
+
+def app():
+    st.write("# Welcome to Ship Detection Application! :satellite:")
+    st.markdown(
+            """
+            This application is build based on YOLOv5 with extral large model. User just
+            upload an image, and press the 'Predict' button to make a prediction base on
+            a training model before.
+
+            ### For more information, please visit:
+
+            - Check out [my github](https://github.com/bills1912)
+            - Jump into YOLOv5 [documentation](https://docs.ultralytics.com/)
+
+        """
+        )
+    ais = pd.read_csv("https://raw.githubusercontent.com/bills1912/marin-vessels-detection/main/data/MarineTraffic_VesselExport_2022-11-25.csv")
+    ais_jakarta = ais[ais['Destination Port'] == 'JAKARTA']
+    ais_list = ais_jakarta.values.tolist()
+    f = folium.Figure(width=1000, height=500)
+    jakarta_vessels = folium.Map(location=[-5.626954250925966, 106.70735731868719], zoom_start=8).add_to(f)
+    ais_data = folium.FeatureGroup(name="marine_vessels")
+    mCluster = MarkerCluster(name="Marine Vessels")
+    for i in ais_list:
+        html = f"<h3>{i[1]}</h3> Vessel Type: {i[8]} </br> Destination Port: {i[2]} </br> Reported Destination: {i[4]} </br> Current Port: {i[6]}\
+                </br> Latitude: {i[10]} </br> Longitude: {i[11]}"
+        iframe = folium.IFrame(html)
+        popup = folium.Popup(iframe, min_width=250, max_width=300)
+        ais_data.add_child(mCluster.add_child(folium.Marker(location=[i[10], i[11]], popup=popup, icon=folium.Icon(color="black", icon="ship", prefix="fa"))))
+        jakarta_vessels.add_child(ais_data)
+    folium_static(jakarta_vessels, width=1370, height=700)

apps/main_model.py

@@ -0,0 +1,32 @@
+import time
+from turtle import width
+import torch
+import numpy as np
+import streamlit as st
+
+def app():
+    st.write("## Ship Imagery Prediction")
+    st.write("### Model evaluation:")
+    eval_col1, eval_col2, eval_col3, eval_col4, eval_col5 = st.columns(spec=5)
+    eval_col1.metric("Precision", "89.52%")
+    eval_col2.metric("Recall", "83.54%")
+    eval_col3.metric("F1-Score", "86.43%")
+    eval_col4.metric("mAP 0.5", "85.39%")
+    eval_col5.metric("mAP 0.5:0.95", "62.63%")
+
+    uploaded_file = st.file_uploader("Choose a ship imagery")
+    if uploaded_file is not None:
+        st.image(uploaded_file, caption='Image to predict')
+        folder_path = st.text_input("Image path",
+                                    help="This field the image path field that the model will predict the object inside the image that we have uploaded",
+                                    placeholder="Copy the path of image to this field")
+
+    prediction = st.button("Predict")
+    if prediction:
+        ship_model = torch.hub.load('ultralytics/yolov5', 'custom', path="apps/model/main_model.pt", force_reload=True)
+        results = ship_model(f"{folder_path}")
+        with st.spinner("Loading..."):
+            time.sleep(3.5)
+            st.success("Done!")
+        st.image(np.squeeze(results.render()))
+        results.print()

multiapp.py

@@ -0,0 +1,80 @@
+"""Frameworks for running multiple Streamlit applications as a single app.
+"""
+import streamlit as st
+
+# app_state = st.experimental_get_query_params()
+# app_state = {k: v[0] if isinstance(v, list) else v for k, v in app_state.items()} # fetch the first item in each query string as we don't have multiple values for each query string key in this example
+
+
+class MultiApp:
+    """Framework for combining multiple streamlit applications.
+    Usage:
+        def foo():
+            st.title("Hello Foo")
+        def bar():
+            st.title("Hello Bar")
+        app = MultiApp()
+        app.add_app("Foo", foo)
+        app.add_app("Bar", bar)
+        app.run()
+    It is also possible keep each application in a separate file.
+        import foo
+        import bar
+        app = MultiApp()
+        app.add_app("Foo", foo.app)
+        app.add_app("Bar", bar.app)
+        app.run()
+    """
+
+    def __init__(self):
+        self.apps = []
+
+    def add_app(self, title, func):
+        """Adds a new application.
+        Parameters
+        ----------
+        func:
+            the python function to render this app.
+        title:
+            title of the app. Appears in the dropdown in the sidebar.
+        """
+        self.apps.append({"title": title, "function": func})
+
+    def run(self):
+        app_state = st.experimental_get_query_params()
+        app_state = {
+            k: v[0] if isinstance(v, list) else v for k, v in app_state.items()
+        }  # fetch the first item in each query string as we don't have multiple values for each query string key in this example
+
+        # st.write('before', app_state)
+
+        titles = [a["title"] for a in self.apps]
+        functions = [a["function"] for a in self.apps]
+        default_radio = titles.index(app_state["page"]) if "page" in app_state else 0
+
+        st.sidebar.title("Navigation")
+
+        title = st.sidebar.radio("Go To", titles, index=default_radio, key="radio")
+
+        app_state["page"] = st.session_state.radio
+        # st.write('after', app_state)
+
+        st.experimental_set_query_params(**app_state)
+        functions[titles.index(title)]()
+
+        st.sidebar.title("Contribute")
+        st.sidebar.info(
+            "This is an open source project and you are very welcome to contribute your "
+            "comments, questions, resources and apps as "
+            "[issues](https://github.com/giswqs/streamlit-geospatial/issues) or "
+            "[pull requests](https://github.com/giswqs/streamlit-geospatial/pulls) "
+            "to the [source code](https://github.com/giswqs/streamlit-geospatial). "
+        )
+        st.sidebar.title("About")
+        st.sidebar.info(
+            """
+            This web [app](https://share.streamlit.io/giswqs/streamlit-geospatial/app.py) is maintained by [Qiusheng Wu](https://wetlands.io). You can follow me on social media:
+             [GitHub](https://github.com/giswqs) | [Twitter](https://twitter.com/giswqs) | [YouTube](https://www.youtube.com/c/QiushengWu) | [LinkedIn](https://www.linkedin.com/in/qiushengwu).
+            This web app URL: <https://streamlit.gishub.org>
+        """
+        )