feat: add app

requirements.txt

@@ -1,3 +1,8 @@
 altair
 pandas
-streamlit
+streamlit
+ultralytics
+pillow
+numpy
+torch
+numpy<2

src/streamlit_app.py

@@ -1,40 +1,161 @@
-import altair as alt
-import numpy as np
-import pandas as pd
 import streamlit as st
+from ultralytics import YOLO
+from PIL import Image
+import numpy as np
+
+# --- CONFIGURATION ---
+MODEL_PATH = "best.pt"
+
+# --- KNOWLEDGE BASE ---
+SYMPTOM_QUESTIONS = {
+    "G1": "Daun berwana putih kecoklatan memanjang seperti mengering pada bagian tepi daun?",
+    "G2": "Pada pagi hari, ditemukan cairan bakteri (seperti butiran air) pada bagian terinfeksi?",
+    "G3": "Luka terlihat seperti anak panah di antara urat daun?",
+    "G4": "Luka berwarna kuning kecoklatan?",
+    "G5": "Pada daun terlihat transparan bila dihadapkan cahaya matahari?",
+    "G6": "Bercak berbentuk belah ketupat (Diamond shape)?",
+    "G7": "Bagian tengah bercak abu-abu/putih, tepi berwarna kecoklatan?",
+    "G8": "Infeksi pada malai/leher berwarna abu-abu?",
+    "G13": "Tanaman kerdil dan daun berubah warna (hijau -> jingga/kemerahan)?",
+    "G025": "Daun menguning, menggulung, mengering dan menjadi layu?",
+    "G026": "Bibit menjadi layu (kresek) tapi sulit dicabut?",
+    "G027": "Warna luka bercak jingga kekuningan dari ujung ke pangkal?",
+    "G028": "Ada bulatan kecil berwarna kuning pada pelepah daun?"
+}
+
+YOLO_PROMPTS = {
+    "bacterial_leaf_blight": ["G1", "G2"], 
+    "bacterial_leaf_streak": ["G3", "G4", "G5"],
+    "blast": ["G6", "G7", "G8"], 
+    "tungro": ["G13"],
+    "brown_spot": [], 
+    "downy_mildew": [],
+    "dead_heart": [],
+    "hispa": [],
+    "normal": []
+}
+
+PRODUCTION_RULES = [
+    {"disease": "Bacterial Leaf Blight (Hawar Daun Bakteri)", "symptoms": ["G1", "G2"], "source": "Literatur 1"},
+    {"disease": "Bacterial Leaf Streak (Bakteri Daun Bergaris)", "symptoms": ["G3", "G4", "G5"], "source": "Literatur 1"},
+    {"disease": "Blast (Blas)", "symptoms": ["G6", "G7"], "source": "Literatur 1"},
+    {"disease": "Neck Blast (Blas Leher)", "symptoms": ["G8"], "source": "Literatur 1"},
+    {"disease": "Tungro", "symptoms": ["G13"], "source": "Literatur 1"},
+    {"disease": "Bacterial Leaf Blight (Alt. Definition)", "symptoms": ["G025", "G026", "G027", "G028"], "source": "Literatur 2"}
+]
+
+# --- MODEL LOADING (Cached so it doesn't reload every click) ---
+@st.cache_resource
+def load_model():
+    try:
+        return YOLO(MODEL_PATH)
+    except Exception as e:
+        st.error(f"Error loading model: {e}")
+        return None
+
+model = load_model()
+
+# --- APP LAYOUT ---
+st.title("🌾 Rice Doctor: Hybrid AI System")
+st.markdown("""
+1. **AI Inference**: Detects disease from image.
+2. **Expert System**: Verifies diagnosis using Forward Chaining logic.
+""")
+
+# Initialize Session State to keep data across re-runs
+if 'prediction' not in st.session_state:
+    st.session_state['prediction'] = None
+if 'priority_codes' not in st.session_state:
+    st.session_state['priority_codes'] = []
+if 'analyzed' not in st.session_state:
+    st.session_state['analyzed'] = False
+
+# --- STEP 1: UPLOAD AND ANALYZE ---
+uploaded_file = st.file_uploader("Step 1: Upload Leaf Image", type=["jpg", "png", "jpeg"])
+
+if uploaded_file is not None:
+    # Display image
+    image = Image.open(uploaded_file)
+    st.image(image, caption="Uploaded Image", use_container_width=True)
+    
+    if st.button("Analyze & Generate Questions", type="primary"):
+        if model:
+            with st.spinner("Analyzing..."):
+                # Run Inference
+                results = model.predict(image, imgsz=640)
+                raw_name = results[0].names[results[0].probs.top1]
+                top_class_name = raw_name.lower().strip()
+                
+                # Save to session state
+                st.session_state['prediction'] = top_class_name
+                st.session_state['priority_codes'] = YOLO_PROMPTS.get(top_class_name, [])
+                st.session_state['analyzed'] = True
+                
+                # Force rerun to show step 2
+                st.rerun()
+
+# --- STEP 2: DYNAMIC QUESTIONS ---
+if st.session_state['analyzed']:
+    st.divider()
+    st.subheader("Step 2: Symptom Verification")
+    
+    pred = st.session_state['prediction']
+    codes = st.session_state['priority_codes']
+    
+    st.info(f"🔍 **AI Prediction:** `{pred}`")
+    
+    if not codes:
+        st.warning("⚠️ No specific verification questions defined for this disease. Please check the **Full Knowledge Base** below.")
+    else:
+        st.write("Please verify this diagnosis by answering the specific symptoms below:")
+    
+    # Dynamic Yes/No Questions
+    user_answers = {}
+    for code in codes:
+        question = SYMPTOM_QUESTIONS.get(code, "Unknown")
+        # Unique key is important in Streamlit
+        ans = st.radio(
+            f"**({code})** {question}", 
+            options=["No", "Yes"], 
+            index=0, # Default to No
+            key=f"q_{code}"
+        )
+        if ans == "Yes":
+            user_answers[code] = True
+
+    st.divider()
+    st.markdown("**Are there other symptoms?** (If the AI missed something, check it here)")
+    
+    # Full Checklist
+    all_symptoms_list = [f"{k}: {v}" for k,v in SYMPTOM_QUESTIONS.items()]
+    other_checks = st.multiselect("Full Knowledge Base", all_symptoms_list)
+    
+    # Process "Other" inputs
+    for item in other_checks:
+        code = item.split(":")[0].strip()
+        user_answers[code] = True
 
-"""
-# Welcome to Streamlit!
-
-Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:.
-If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community
-forums](https://discuss.streamlit.io).
-
-In the meantime, below is an example of what you can do with just a few lines of code:
-"""
-
-num_points = st.slider("Number of points in spiral", 1, 10000, 1100)
-num_turns = st.slider("Number of turns in spiral", 1, 300, 31)
-
-indices = np.linspace(0, 1, num_points)
-theta = 2 * np.pi * num_turns * indices
-radius = indices
-
-x = radius * np.cos(theta)
-y = radius * np.sin(theta)
-
-df = pd.DataFrame({
-    "x": x,
-    "y": y,
-    "idx": indices,
-    "rand": np.random.randn(num_points),
-})
-
-st.altair_chart(alt.Chart(df, height=700, width=700)
-    .mark_point(filled=True)
-    .encode(
-        x=alt.X("x", axis=None),
-        y=alt.Y("y", axis=None),
-        color=alt.Color("idx", legend=None, scale=alt.Scale()),
-        size=alt.Size("rand", legend=None, scale=alt.Scale(range=[1, 150])),
-    ))
+    # --- STEP 3: FORWARD CHAINING ---
+    if st.button("Step 3: Run Forward Chaining", type="primary"):
+        user_facts = set(user_answers.keys())
+        
+        if not user_facts:
+            st.error("⚠️ **Inconclusive:** You selected 'No' for everything.")
+        else:
+            matches = []
+            for rule in PRODUCTION_RULES:
+                required = set(rule['symptoms'])
+                if required.issubset(user_facts):
+                    matches.append(rule)
+            
+            if matches:
+                st.success("✅ **Disease Confirmed!**")
+                for m in matches:
+                    st.markdown(f"""
+                    ### {m['disease']}
+                    - **Source:** {m['source']}
+                    - **Logic:** Found all required symptoms {m['symptoms']}
+                    """)
+            else:
+                st.error("❌ **No Exact Match Found**")
+                st.write("The symptoms you selected do not perfectly fit any strict rule in the knowledge base.")