Beautification and add info

.streamlit/config.toml

@@ -162,7 +162,7 @@ enableWebsocketCompression = false
 # Enable serving files from a `static` directory in the running app's
 # directory.
 # Default: false
-enableStaticServing = false
+enableStaticServing = true
 
 # Server certificate file for connecting via HTTPS.
 # Must be set at the same time as "server.sslKeyFile".

README.md

@@ -1,6 +1,6 @@
 ---
 title: Tree Species Classification
-emoji: 🔥
+emoji: 🌲
 colorFrom: pink
 colorTo: green
 sdk: streamlit
@@ -10,4 +10,4 @@ pinned: false
 short_description: Individual tree species identification
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+This project predicts the tree species, given its point cloud data, and also calculates the tree inventory such as tree height, canopy volume and DBH.

app.py

@@ -18,32 +18,106 @@ with st.spinner("Loading PointNet++ model..."):
     classifier = pn2.get_model(num_class=4, normal_channel=False)
     classifier.load_state_dict(checkpoint['model_state_dict'])
     classifier.eval()
+    
+st.sidebar.markdown(
+    body=
+        "<div style='text-align: justify;'>The species <strong>Pinus sylvestris "
+        "(Scots Pine), Fagus sylvatica (European Beech), Picea abies (Norway Spruce), "
+        "and Betula pendula (Silver Birch)</strong> are native to Europe and parts "
+        "of Asia but are also found in India (Parts of Himachal Pradesh, "
+        "Uttarakhand, Jammu and Kashmir, Sikkim and Arunachal Pradesh). "
+        "These temperate species, typically thriving in boreal and montane ecosystems, "
+        "are occasionally introduced in cooler Indian regions like the Himalayan "
+        "foothills for afforestation or experimental forestry, where climatic "
+        "conditions are favourable. However, their growth and ecological interactions "
+        "in India may vary significantly due to the region's unique biodiversity "
+        "and environmental factors.<br><br>"
+        "This AI-powered application employs the PointNet++ deep learning "
+        "architecture, optimized for processing 3D point cloud data from "
+        "individual <code>.laz</code> files (fused aerial and terrestrial LiDAR) "
+        "to classify tree species up to four classes (<strong>Pinus sylvestris, "
+        "Fagus sylvatica, Picea abies, and Betula pendula</strong>) "
+        "with associated confidence scores. Additionally, it calculates critical "
+        "metrics such as Diameter at Breast Height (DBH), actual height and "
+        "customizable canopy volume, enabling precise refinement of predictions "
+        "and analyses. By integrating species-specific and volumetric insights, "
+        "the tool enhances ecological research workflows, facilitating data-driven "
+        "decision-making.</div>"
+        ,
+    unsafe_allow_html=True,
+)
+st.markdown(
+    """
+    <style>
+    [data-testid="stSidebar"] {
+        background-image: url("static/sidebar.png");
+        background-size: cover;
+        background-position: center;
+    }
+    </style>
+    """,
+    unsafe_allow_html=True
+)
 
-st.title("Tree Species Identification")
+st.header("ArborSphere")
+st.subheader("Tree Identity and Biometrics")
 
 uploaded_file = st.file_uploader(
     label="Upload Point Cloud Data", 
     type=['laz', 'las', 'pcd'], 
     help="Please upload trees with ground points removed"
 )
-Z_THRESHOLD = st.slider(
-    label="Z-Threshold(%)", 
-    min_value=10, 
-    max_value=90, 
-    value=50, 
-    step=1, 
-    help="Please select a Z-Threshold for canopy volume calculation"
-)
-DBH_HEIGHT = st.slider(
-    label="DBH Height(m)", 
-    min_value=1.3,
-    max_value=1.4,
-    value=1.4,
-    step=0.01,
-    help="Enter height used for DBH calculation"
-)
-proceed = None
 
+col1, col2 = st.columns([2, 2])
+with col1:
+    st.image("static/canopy_info.jpg")
+with col2:
+    CANOPY_VOLUME = st.slider(
+        label="Canopy Volume in % (Z)", 
+        min_value=10, 
+        max_value=90, 
+        value=70, 
+        step=1, 
+        help=
+            "Adjust the Z-threshold value to calculate the canopy volume " 
+            "within specified limits, it uses Quickhull and DBSCAN algorithms. "
+            
+            "The Quickhull algorithm computes the convex hull of a set of points "
+            "by identifying extreme points to form an initial boundary and recursively "
+            "refining it by adding the farthest points until all points lie within the "
+            "convex boundary. It uses a divide-and-conquer approach, similar to QuickSort. "
+            
+            "DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a "
+            "density-based clustering algorithm that groups densely packed points within "
+            "a specified distance 'eps' and minimum points 'minpoints', while treating "
+            "sparse points as noise. It effectively identifies arbitrarily shaped clusters "
+            "and handles outliers, making it suitable for spatial data and anomaly detection."
+    )
+    
+col1, col2 = st.columns([2, 2])
+with col1:
+    st.image("static/DBH_info.jpg")
+with col2:
+    DBH_HEIGHT = st.slider(
+        label="DBH (Diameter above Breast Height, in metres) (H)", 
+        min_value=1.3,
+        max_value=1.4,
+        value=1.4,
+        step=0.01,
+        help=
+            "Adjust to calculate the DBH value within specified limits, "
+            "it utilizes Least square circle fitting method Levenberg-Marquardt "
+            "optimization technique."
+            
+            "The Least Squares Circle Fitting method is used to find the "
+            "best-fitting circle to a set of 2D points by minimizing the sum of "
+            "squared distances between each point and the circle's circumference."
+            "Levenberg-Marquardt Optimization is used to fit models (like circles) "
+            "to point cloud data by minimizing the error between the model and the "
+            "actual points."
+    )
+    
+proceed = None
 if uploaded_file:
     try:
         with st.spinner("Reading point cloud file..."):
@@ -72,7 +146,7 @@ if proceed:
             z_max = np.max(points[:, 2])
             height = z_max - z_min
             
-            canopy_volume, canopy_points = calc_canopy_volume(points, Z_THRESHOLD, height, z_min)
+            canopy_volume, canopy_points = calc_canopy_volume(points, CANOPY_VOLUME, height, z_min)
                 
         with st.spinner("Visualizing point cloud..."):
             fig = go.Figure()
@@ -120,9 +194,27 @@ if proceed:
                     yaxis_title="Y",
                     zaxis_title="Z",
                     aspectmode='data'
-                )
+                ),
+                showlegend=False
             )
-            st.plotly_chart(fig, use_container_width=True)
+            col1, col2, col3 = st.columns([1, 3, 1])
+            with col2:
+                st.markdown("""
+                <style>
+                .centered-plot {
+                    text-align: center;
+                }
+                </style>
+                """, unsafe_allow_html=True)
+                st.plotly_chart(fig, use_container_width=True)
+            hide_st_style = """
+                <style>
+                #MainMenu {visibility: hidden;}
+                footer {visibility: hidden;}
+                header {visibility: hidden;}
+                </style>
+            """
+            st.markdown(hide_st_style, unsafe_allow_html=True)
             
             
         with st.spinner("Running inference..."):