Create app.py

app.py

@@ -0,0 +1,119 @@
+import streamlit as st
+import pandas as pd
+import pytesseract
+from PIL import Image
+from rapidfuzz import fuzz, utils
+import io
+
+# --- CONFIGURATION ---
+st.set_page_config(page_title="AI Student Grader", layout="wide")
+st.title("📝 AI Student Answer Grader")
+st.markdown("Upload answer sheets and an answer key to automatically calculate marks.")
+
+# --- SIDEBAR SETTINGS ---
+st.sidebar.header("Grading Settings")
+accuracy_threshold = st.sidebar.slider("Minimum Accuracy Threshold (%)", 0, 100, 70)
+marks_per_question = st.sidebar.number_input("Marks per correct answer", value=1.0)
+
+# --- HELPER FUNCTIONS ---
+def perform_ocr(image):
+    """Extracts text from an uploaded image."""
+    img = Image.open(image)
+    # Optional: Add image preprocessing here (grayscale, thresholding)
+    text = pytesseract.image_to_string(img)
+    return text.strip()
+
+def compare_answers(student_text, answer_key, threshold):
+    """
+    Compares student text with answer key using Fuzzy Matching.
+    DeepSeek-R1 style logic: We look for the presence of key concepts.
+    """
+    # Simple line-by-line comparison (assuming 1 question per line or similar structure)
+    # For complex papers, you'd split by question numbers
+    score = 0
+    key_lines = [line.strip() for line in answer_key.split('\n') if line.strip()]
+    student_lines = [line.strip() for line in student_text.split('\n') if line.strip()]
+    
+    details = []
+    
+    for i, correct_ans in enumerate(key_lines):
+        match_found = False
+        highest_match = 0
+        
+        # Compare against each line in student text to find the best match for this answer
+        for s_line in student_lines:
+            similarity = fuzz.token_set_ratio(correct_ans, s_line)
+            if similarity > highest_match:
+                highest_match = similarity
+        
+        if highest_match >= threshold:
+            score += marks_per_question
+            match_found = True
+            
+        details.append({
+            "Question": i + 1,
+            "Match %": round(highest_match, 2),
+            "Status": "Correct" if match_found else "Incorrect"
+        })
+            
+    return score, details
+
+# --- UI LAYOUT ---
+col1, col2 = st.columns(2)
+
+with col1:
+    st.subheader("1. Reference Answer Key")
+    key_input_type = st.radio("Key Format", ["Text Input", "Upload Image"])
+    
+    if key_input_type == "Text Input":
+        answer_key_text = st.text_area("Paste the correct answers (one per line):")
+    else:
+        key_img = st.file_uploader("Upload Answer Key Image", type=['png', 'jpg', 'jpeg'])
+        if key_img:
+            answer_key_text = perform_ocr(key_img)
+            st.text_area("Extracted Key (Edit if needed):", value=answer_key_text)
+
+with col2:
+    st.subheader("2. Student Answer Sheets")
+    student_images = st.file_uploader("Upload Student Images (Max 5)", type=['png', 'jpg', 'jpeg'], accept_multiple_files=True)
+
+# --- PROCESSING ---
+if st.button("Calculate Marks"):
+    if not answer_key_text or not student_images:
+        st.error("Please provide both the answer key and student images.")
+    else:
+        results = []
+        
+        progress_bar = st.progress(0)
+        for idx, img_file in enumerate(student_images):
+            # 1. OCR
+            extracted_text = perform_ocr(img_file)
+            
+            # 2. Compare
+            score, details = compare_answers(extracted_text, answer_key_text, accuracy_threshold)
+            
+            # 3. Store Results
+            results.append({
+                "Student Name": img_file.name.split('.')[0], # Uses filename as name
+                "Raw Score": score,
+                "Final Marks": f"{score}/{len(answer_key_text.splitlines()) * marks_per_question}",
+                "Match Percentage": f"{accuracy_threshold}%"
+            })
+            progress_bar.progress((idx + 1) / len(student_images))
+
+        # --- DISPLAY RESULTS ---
+        df = pd.DataFrame(results)
+        st.subheader("📊 Results Overview")
+        st.table(df)
+
+        # --- EXCEL EXPORT ---
+        output = io.BytesIO()
+        with pd.ExcelWriter(output, engine='openpyxl') as writer:
+            df.to_excel(writer, index=False, sheet_name='Grades')
+        
+        st.download_button(
+            label="📥 Download Excel Sheet",
+            data=output.getvalue(),
+            file_name="student_grades.xlsx",
+            mime="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+        )