Upload 2 files

app.py

@@ -0,0 +1,253 @@
+import streamlit as st
+from transformers import pipeline
+import networkx as nx
+import matplotlib.pyplot as plt
+from io import BytesIO
+import torch
+from diffusers import StableDiffusionPipeline
+from PIL import Image
+
+# 1. Page Configuration
+st.set_page_config(
+    page_title="GreenAI - Environmental Intelligence Dashboard",
+    layout="wide",
+    initial_sidebar_state="expanded"
+)
+
+# Custom CSS for a modern look (optional, but enhances aesthetics)
+st.markdown("""
+    <style>
+    .stApp {
+        background-color: #f0f2f6;
+        color: #333333;
+    }
+    .stTabs [data-baseweb="tab-list"] button [data-testid="stMarkdownContainer"] p {
+        font-size: 1.1rem;
+        font-weight: bold;
+    }
+    h1, h2, h3, h4, h5, h6 {
+        color: #0068c9;
+    }
+    .stButton>button {
+        background-color: #4CAF50;
+        color: white;
+        font-size: 1rem;
+        padding: 10px 20px;
+        border-radius: 8px;
+        border: none;
+        box-shadow: 0 4px 8px 0 rgba(0,0,0,0.2);
+        transition: 0.3s;
+    }
+    .stButton>button:hover {
+        background-color: #45a049;
+        box-shadow: 0 8px 16px 0 rgba(0,0,0,0.2);
+    }
+    .stTextInput>div>div>input {
+        border-radius: 8px;
+        border: 1px solid #ccc;
+        padding: 10px;
+    }
+    .stTextArea>div>div>textarea {
+        border-radius: 8px;
+        border: 1px solid #ccc;
+        padding: 10px;
+    }
+    </style>
+""", unsafe_allow_html=True)
+
+st.title("🌱 GreenAI: Smart Environmental AI Assistant")
+st.markdown("Detects pollution types, generates visuals, fills blanks, and maps entities.")
+
+# 2. Load models
+@st.cache_resource
+def load_models():
+    with st.spinner("🔄 Loading AI models... This might take a moment."):
+        classifier = pipeline("zero-shot-classification", model="facebook/bart-large-mnli")
+        ner = pipeline("ner", model="dslim/bert-base-NER", grouped_entities=True)
+        fill_mask = pipeline("fill-mask", model="roberta-base")
+
+        # Stable Diffusion via diffusers
+        if torch.cuda.is_available():
+            image_gen = StableDiffusionPipeline.from_pretrained(
+                "runwayml/stable-diffusion-v1-5",
+                torch_dtype=torch.float16,
+            ).to("cuda")
+        else:
+            image_gen = StableDiffusionPipeline.from_pretrained(
+                "runwayml/stable-diffusion-v1-5"
+            ).to("cpu")
+    st.success("✅ All models loaded successfully!")
+    return classifier, ner, fill_mask, image_gen
+
+classifier, ner, fill_mask, image_gen = load_models()
+
+# 3. Define functions
+def classify_text(text):
+    labels = ["Waste Management", "Water Management", "Air Pollution", "Recycling", "Energy Conservation"]
+    result = classifier(text, candidate_labels=labels)
+    return f"**Label:** {result['labels'][0]}, **Score:** {result['scores'][0]:.2f}"
+
+def generate_image(prompt):
+    # Add a loading spinner for image generation
+    with st.spinner("🎨 Generating image... This can take up to a minute."):
+        image = image_gen(prompt).images[0]
+    return image
+
+ENV_TERMS = [
+    # Pollution
+    "pollution", "air", "water", "soil", "noise", "radiation", "smog", "contamination",
+    "runoff", "eutrophication", "emission", "discharge",
+
+    # Waste
+    "waste", "garbage", "litter", "plastic", "microplastic", "sewage", "industrial",
+    "landfill", "toxic", "hazardous", "e-waste", "compost", "recyclables",
+
+    # Natural Resources
+    "river", "ocean", "lake", "forest", "biodiversity", "wildlife", "wetland",
+    "coral", "marine", "vegetation", "habitat",
+
+    # Climate
+    "climate", "warming", "greenhouse", "carbon", "dioxide", "methane", "ozone",
+    "temperature", "drought", "flood", "acid rain", "co2", "ghg",
+
+    # Sustainability
+    "recycle", "reuse", "compost", "conservation", "renewable", "solar", "wind",
+    "hydro", "clean energy", "green tech", "sustainable", "biodegradable",
+    "energy efficiency", "net zero",
+
+    # Industrial
+    "factory", "industry", "agriculture", "vehicle", "transport", "plant",
+    "chemical", "refinery", "pesticide", "fertilizer", "manufacturing"
+]
+
+def ner_with_graph(text):
+    entities = ner(text)
+    G = nx.Graph()
+
+    seen_words = set()
+
+    # Add NER entities
+    for ent in entities:
+        raw_label = ent["entity_group"]
+        clean_label = {
+            "PER": "Person", "LOC": "Location", "ORG": "Organization", "MISC": "Misc"
+        }.get(raw_label, raw_label)
+
+        word = ent["word"].replace("##", "").strip()
+        if word not in seen_words:
+            G.add_node(word, label=clean_label)
+            seen_words.add(word)
+
+    # Add environmental terms
+    words = text.split()
+    for word in words:
+        clean_word = word.lower().strip(",.")
+        if clean_word in ENV_TERMS:
+            if word not in seen_words:
+                G.add_node(word, label="Environmental")
+                seen_words.add(word)
+
+    # Link nodes in order of appearance
+    all_nodes = list(G.nodes()) # Use G.nodes() after adding all nodes
+    for i in range(len(all_nodes) - 1):
+        G.add_edge(all_nodes[i], all_nodes[i + 1])
+
+    # Draw the graph
+    fig, ax = plt.subplots(figsize=(10, 7)) # Increased figure size for better readability
+    pos = nx.spring_layout(G, seed=42, k=0.7) # Adjusted k for better spacing
+
+    labels = nx.get_node_attributes(G, "label")
+    color_map = {
+        "Person": "orange", "Location": "skyblue", "Organization": "lightgray",
+        "Misc": "violet", "Environmental": "lightgreen"
+    }
+    node_colors = [color_map.get(labels.get(node, ""), "white") for node in G.nodes]
+
+    # Draw with enhanced styling
+    nx.draw(
+        G, pos, with_labels=False, node_color=node_colors, edge_color="gray",
+        node_size=3000, font_weight="bold", alpha=0.9, linewidths=1.5, edgecolors="black"
+    )
+    # Add labels with background for better visibility
+    for node, (x, y) in pos.items():
+        text_label = f"{node}\n({labels[node]})" if node in labels else node
+        ax.text(x, y, text_label, horizontalalignment='center', verticalalignment='center',
+                fontsize=9, color='black', bbox=dict(facecolor='white', alpha=0.7, edgecolor='none', boxstyle='round,pad=0.3'))
+
+    st.pyplot(fig) # Display the plot directly in Streamlit
+    plt.close(fig) # Close the figure to free up memory
+
+def fill_blank(text):
+    if "<mask>" not in text:
+        st.error("❌ Please include exactly one `<mask>` token in your sentence.")
+        return []
+    try:
+        results = fill_mask(text)
+        return [r["sequence"] for r in results]
+    except Exception as e:
+        st.error(f"Error: {str(e)}")
+        return []
+
+# 4. Streamlit UI
+tab1, tab2, tab3, tab4 = st.tabs([
+    "1️⃣ Sentence Classification",
+    "2️⃣ Image Generation",
+    "3️⃣ NER + Entity Graph",
+    "4️⃣ Fill-in-the-Blank"
+])
+
+with tab1:
+    st.header("Sentence Classification")
+    st.write("Automatically categorize environmental sentences.")
+    text_input = st.text_area("📝 Enter an environmental sentence:", height=100,
+                              placeholder="Example: The factory discharged toxic waste into the river.")
+    if st.button("🚀 Classify Sentence"):
+        if text_input:
+            result = classify_text(text_input)
+            st.info(result)
+        else:
+            st.warning("Please enter some text to classify.")
+
+with tab2:
+    st.header("Image Generation")
+    st.write("Generate environmental images from text prompts.")
+    img_prompt_input = st.text_area("🎨 Prompt for environmental image:", height=100,
+                                    placeholder="Example: A serene forest with clean river and diverse wildlife.")
+    if st.button("🧠 Generate Image"):
+        if img_prompt_input:
+            generated_image = generate_image(img_prompt_input)
+            st.image(generated_image, caption=img_prompt_input, use_column_width=True)
+        else:
+            st.warning("Please enter a prompt to generate an image.")
+
+with tab3:
+    st.header("Named Entity Recognition (NER) & Entity Graph")
+    st.write("Extract and visualize named entities and environmental terms from text.")
+    ner_input_text = st.text_area("🔎 Sentence with named entities:", height=100,
+                                  value="The CEO of EcoCorp announced a new initiative in London to reduce plastic waste.",
+                                  placeholder="Example: John lives in New York and works for Google.")
+    if st.button("📊 Extract & Visualize Entities"):
+        if ner_input_text:
+            st.info("Generating entity graph...")
+            ner_with_graph(ner_input_text)
+        else:
+            st.warning("Please enter some text for entity extraction.")
+
+with tab4:
+    st.header("Fill-in-the-Blank")
+    st.write("Predict missing words in environmental sentences using `<mask>`.")
+    fill_input_text = st.text_area("🧩 Enter sentence with `<mask>`:", height=100,
+                                   value="Air <mask> is harmful to public health.",
+                                   placeholder="Example: The Amazon <mask> is a vital ecosystem.")
+    if st.button("🔍 Predict Missing Word(s)"):
+        if fill_input_text:
+            predictions = fill_blank(fill_input_text)
+            if predictions:
+                st.subheader("📚 Top Predictions:")
+                for i, pred in enumerate(predictions):
+                    st.write(f"{i+1}. {pred}")
+        else:
+            st.warning("Please enter a sentence with `<mask>`.")
+
+st.markdown("---")
+st.markdown("Developed with ❤️ for a greener future.")

requirements.txt

@@ -0,0 +1,7 @@
+streamlit
+transformers
+networkx
+matplotlib
+torch
+diffusers
+accelerate # Highly recommended for faster Stable Diffusion on GPU