Update app.py

app.py

@@ -1,178 +1,93 @@
 import streamlit as st
-import pymupdf  
-import re
-import traceback
-import faiss
-import numpy as np
-import requests
-from rank_bm25 import BM25Okapi
-from sentence_transformers import SentenceTransformer
-from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain_groq import ChatGroq
 import torch
-import os 
-
-os.environ["STREAMLIT_WATCHDOG_TYPE"] = "none"
-
-st.set_page_config(page_title="Financial Insights Chatbot", page_icon="📊", layout="wide")
-
-device = "cuda" if torch.cuda.is_available() else "cpu"
-
-GROQ_API_KEY = os.getenv("GROQ_API_KEY")
-ALPHA_VANTAGE_API_KEY = os.getenv("ALPHA_VANTAGE_API_KEY")
-
-try:
-    llm = ChatGroq(temperature=0, model="llama3-70b-8192", api_key=GROQ_API_KEY)
-    st.success("✅ LLM initialized successfully. Using llama3-70b-8192")
-except Exception as e:
-    st.error("❌ Failed to initialize Groq LLM.")
-    traceback.print_exc()
-
-embedding_model = SentenceTransformer("baconnier/Finance2_embedding_small_en-V1.5", device=device)
-
-text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
-
-def fetch_financial_data(company_ticker):
-    if not company_ticker:
-        return "No ticker symbol provided. Please enter a valid company ticker."
-
-    try:
-        overview_url = f"https://www.alphavantage.co/query?function=OVERVIEW&symbol={company_ticker}&apikey={ALPHA_VANTAGE_API_KEY}"
-        overview_response = requests.get(overview_url)
-
-        if overview_response.status_code == 200:
-            overview_data = overview_response.json()
-            market_cap = overview_data.get("MarketCapitalization", "N/A")
-        else:
-            return "Error fetching company overview."
-
-        income_url = f"https://www.alphavantage.co/query?function=INCOME_STATEMENT&symbol={company_ticker}&apikey={ALPHA_VANTAGE_API_KEY}"
-        income_response = requests.get(income_url)
-
-        if income_response.status_code == 200:
-            income_data = income_response.json()
-            annual_reports = income_data.get("annualReports", [])
-            revenue = annual_reports[0].get("totalRevenue", "N/A") if annual_reports else "N/A"
-        else:
-            return "Error fetching income statement."
-
-        return f"Market Cap: ${market_cap}\nTotal Revenue: ${revenue}"
-
-    except Exception as e:
-        traceback.print_exc()
-        return "Error fetching financial data."
-
-def extract_and_embed_text(pdf_file):
-    """Processes PDFs and generates embeddings with GPU acceleration using pymupdf."""
-    try:
-        docs, tokenized_texts = [], []
-
-        with pymupdf.open(stream=pdf_file.read(), filetype="pdf") as doc:
-            full_text = "\n".join(page.get_text("text") for page in doc)
-            chunks = text_splitter.split_text(full_text)
-            for chunk in chunks:
-                docs.append(chunk)
-                tokenized_texts.append(chunk.split())
-
-        embeddings = embedding_model.encode(docs, batch_size=64, convert_to_numpy=True, normalize_embeddings=True)
-
-        embedding_dim = embeddings.shape[1]
-        index = faiss.IndexHNSWFlat(embedding_dim, 32)
-        index.add(embeddings)
-
-        bm25 = BM25Okapi(tokenized_texts)
-
-        return docs, embeddings, index, bm25
-    except Exception as e:
-        traceback.print_exc()
-        return [], [], None, None
-
-def retrieve_relevant_docs(user_query, docs, index, bm25):
-    """Hybrid search using FAISS cosine similarity & BM25 keyword retrieval."""
-    query_embedding = embedding_model.encode(user_query, convert_to_numpy=True, normalize_embeddings=True)
-    _, faiss_indices = index.search(np.array([query_embedding]), 8)
-    bm25_scores = bm25.get_scores(user_query.split())
-    bm25_indices = np.argsort(bm25_scores)[::-1][:8]
-    combined_indices = list(set(faiss_indices[0]) | set(bm25_indices))
-
-    return [docs[i] for i in combined_indices[:3]]
-
-def generate_response(user_query, pdf_ticker, ai_ticker, mode, uploaded_file):
-    try:
-        if mode == "📄 PDF Upload Mode":
-            docs, embeddings, index, bm25 = extract_and_embed_text(uploaded_file)
-            if not docs:
-                return "❌ Error extracting text from PDF."
-
-            retrieved_docs = retrieve_relevant_docs(user_query, docs, index, bm25)
-            context = "\n\n".join(retrieved_docs)
-    
-            # Avoid using 'None' in prompt
-            prompt = f"Based on the uploaded financial report, answer the following query:\n{user_query}\n\nRelevant context:\n{context}"
-
-
-        elif mode == "🌍 Live Data Mode":
-            financial_info = fetch_financial_data(ai_ticker)
-            prompt = f"Analyze the financial status of {ai_ticker} based on:\n{financial_info}\n\nUser Query: {user_query}"
-        else:
-            return "Invalid mode selected."
-
-        response = llm.invoke(prompt)
-        return response.content
-    except Exception as e:
-        traceback.print_exc()
-        return "Error generating response."
-
-st.markdown(
-    "<h1 style='text-align: center; color: #4CAF50;'> FinQuery RAG Chatbot</h1>", 
-    unsafe_allow_html=True
-)
-st.markdown(
-    "<h5 style='text-align: center; color: #666;'>Analyze financial reports or fetch live financial data effortlessly!</h5>", 
-    unsafe_allow_html=True
-)
-
-col1, col2 = st.columns(2)
-
-with col1:
-    st.markdown("### 🏢 **Choose Your Analysis Mode**")
-    mode = st.radio("", ["📄 PDF Upload Mode", "🌍 Live Data Mode"], horizontal=True)
-
-with col2:
-    st.markdown("###  **Enter Your Query**")
-    user_query = st.text_input("💬 What financial insights are you looking for?")
-
-st.markdown("---")
-uploaded_file, company_ticker = None, None
-
-if mode == "📄 PDF Upload Mode":
-    st.markdown("### 📂 Upload Your Financial Report")
-    uploaded_file = st.file_uploader("🔼 Upload PDF Report", type=["pdf"])
-    company_ticker = None 
-
-else:
-    st.markdown("### 🌍 Live Market Data")
-    company_ticker = st.text_input("🏢 Enter Company Ticker Symbol", placeholder="e.g., AAPL, MSFT")
-    uploaded_file = None
-
-# 🎯 Submit Button
-if st.button("Analyze Now"):
-    if mode == "📄 PDF Upload Mode" and not uploaded_file:
-        st.error("❌ Please upload a PDF file.")
-    elif mode == "🌍 Live Data Mode" and not company_ticker:
-        st.error("❌ Please enter a valid company ticker symbol.")
-    else:
-        with st.spinner(" Your Query is Processing, this can take up to 5 - 7 minutes ⏳"):
-            if mode == "📄 PDF Upload Mode":
-                response = generate_response(user_query, company_ticker, None, mode, uploaded_file)
-            else:
-                response = generate_response(user_query, None, company_ticker, mode, uploaded_file)
-
-            st.markdown("---")
-            st.markdown("<h3 style='color: #4CAF50;'>💡 AI Response</h3>", unsafe_allow_html=True)
-            st.write(response)
-
-
-# 📌 Footer
-st.markdown("---")
-
+import torch.nn as nn
+import timm
+import numpy as np
+import cv2
+from PIL import Image
+import io
+
+# Hide Streamlit warnings and UI elements
+st.set_page_config(layout="wide")
+st.markdown("""
+    <style>
+        footer {visibility: hidden;}
+    </style>
+""", unsafe_allow_html=True)
+
+# === Model Definition ===
+class MobileViTSegmentation(nn.Module):
+    def __init__(self, encoder_name='mobilevit_s', pretrained=False):
+        super().__init__()
+        self.backbone = timm.create_model(encoder_name, features_only=True, pretrained=pretrained)
+        self.encoder_channels = self.backbone.feature_info.channels()
+
+        self.decoder = nn.Sequential(
+            nn.Conv2d(self.encoder_channels[-1], 128, kernel_size=3, padding=1),
+            nn.Upsample(scale_factor=2, mode='bilinear'),
+            nn.Conv2d(128, 64, kernel_size=3, padding=1),
+            nn.Upsample(scale_factor=2, mode='bilinear'),
+            nn.Conv2d(64, 32, kernel_size=3, padding=1),
+            nn.Upsample(scale_factor=2, mode='bilinear'),
+            nn.Conv2d(32, 1, kernel_size=1),
+            nn.Sigmoid()
+        )
+
+    def forward(self, x):
+        feats = self.backbone(x)
+        out = self.decoder(feats[-1])
+        out = nn.functional.interpolate(out, size=(x.shape[2], x.shape[3]), mode='bilinear', align_corners=False)
+        return out
+
+# === Load Model ===
+@st.cache_resource
+def load_model():
+    model = MobileViTSegmentation()
+    state_dict = torch.load("mobilevit_teeth_segmentation.pth", map_location="cpu")
+    model.load_state_dict(state_dict)
+    model.eval()
+    return model
+
+model = load_model()
+
+# === Preprocessing ===
+def preprocess_image(image: Image.Image):
+    image = image.convert("RGB").resize((256, 256))
+    arr = np.array(image).astype(np.float32) / 255.0
+    arr = np.transpose(arr, (2, 0, 1))  # HWC → CHW
+    tensor = torch.tensor(arr).unsqueeze(0)  # Add batch dim
+    return tensor
+
+# === Postprocessing: Overlay Mask ===
+def overlay_mask(image_pil, mask_tensor, threshold=0.7):
+    image = np.array(image_pil.resize((256, 256)))
+    mask = mask_tensor.squeeze().detach().numpy()
+    mask_bin = (mask > threshold).astype(np.uint8) * 255
+
+    mask_color = np.zeros_like(image)
+    mask_color[..., 2] = mask_bin  # Blue mask
+
+    overlayed = cv2.addWeighted(image, 1.0, mask_color, 0.5, 0)
+    return overlayed
+
+# === UI ===
+st.title("🦷 Tooth Segmentation with MobileViT")
+st.write("Upload an image to segment the **visible teeth area** using a lightweight MobileViT segmentation model.")
+
+uploaded_file = st.file_uploader("Upload an Image", type=["jpg", "jpeg", "png"])
+
+if uploaded_file:
+    image = Image.open(uploaded_file)
+    tensor = preprocess_image(image)
+
+    with st.spinner("Segmenting..."):
+        with torch.no_grad():
+            pred = model(tensor)[0]
+
+    overlayed_img = overlay_mask(image, pred)
+
+    col1, col2 = st.columns(2)
+    with col1:
+        st.image(image, caption="Original Image", use_container_width=True)
+    with col2:
+        st.image(overlayed_img, caption="Tooth Mask Overlay", use_container_width=True)