YaazhiniS/OMR-Evaluation

🎯 Automated OMR Evaluation & Scoring System

An automated OMR (Optical Mark Recognition) evaluation system built for Innomatics Research Labs to process placement readiness assessments efficiently.

📋 Problem Statement

At Innomatics Research Labs, we conduct placement readiness assessments across roles like Data Analytics and AI/ML for Data Science with Generative AI courses. Each exam uses standardized OMR sheets with 100 questions distributed as 20 per subject across 5 subjects.

The manual evaluation process was:

• —⏰ Time-consuming (delays in releasing results)
• —❌ Error-prone (human miscounts)
• —💰 Resource-intensive (requires multiple evaluators)

🚀 Solution

This automated system provides:

• —📱 Mobile phone camera capture support
• —🔄 Automatic image preprocessing and perspective correction
• —🎯 Accurate bubble detection using Computer Vision
• —📊 Subject-wise scoring and total score calculation
• —🌐 Web-based interface for easy evaluation
• —📈 Real-time results and statistics
• —💾 Secure database storage with audit trail

🛠️ Tech Stack

Core OMR Processing

• —Python - Primary programming language
• —OpenCV - Image preprocessing, bubble detection, perspective correction
• —NumPy - Image array manipulation and calculations
• —Pandas - Data processing and Excel handling
• —SQLite - Database for storing results

Web Application

• —Streamlit - Frontend web interface
• —Pillow - Image manipulation and format conversion

📦 Installation

1. 1.Clone the repository:

git clone https://github.com/Yaazhini25/omr-evaluation-system.git
cd omr-evaluation-system

1. 1.Install dependencies:

pip install -r requirements.txt

1. 1.Run the application:

streamlit run app.py

🎮 Usage

Step 1: Prepare Answer Key

• —Create an Excel file with subjects as columns (Python, EDA, SQL, POWER BI, Statistics)
• —Each column should contain 20 correct answers in format "1 - a", "2 - b", etc.
• —Use letters a, b, c, d for the 4 answer choices

Step 2: Upload Answer Key

• —Click "Browse files" and select your Excel answer key
• —Verify the preview shows correct subjects and answers

Step 3: Upload OMR Sheets

• —Take clear, well-lit photos of completed OMR sheets
• —Ensure sheets are flat with all bubbles clearly visible
• —Upload one or multiple images (PNG, JPG, JPEG)

Step 4: Evaluate

• —Optionally enter student name
• —Click "Start Evaluation"
• —Review results and download CSV reports

📊 Features

• —✅ <0.5% Error Tolerance - Meets Innomatics quality standards
• —⚡ Minutes vs Days - Reduces evaluation time from days to minutes
• —🔍 Debug Mode - Shows processing steps for troubleshooting
• —📈 Real-time Statistics - Subject-wise averages and performance metrics
• —💾 Data Persistence - All results stored in SQLite database
• —📱 Mobile Compatible - Works with smartphone camera captures
• —🌐 Web Interface - Easy-to-use browser-based evaluation

🗂️ Project Structure

omr-evaluation-system/
├── app.py                      # Main Streamlit application
├── omr_preprocessing.py        # Image preprocessing functions
├── omr_bubble_detection.py     # Bubble detection algorithms
├── omr_scoring.py             # Scoring and answer key processing
├── omr_results.db             # Backend DB to view the results
├── ans_key.py                 # Load the answer key excel file
├── db_setup.py                # Database operations
├── db_drop.py                 # Drop old database
├── db_checkup.py              # To verify database on the backend 
├── requirements.txt           # Python dependencies
├── Dockerfile                 # Docker configuration
├── README.md
├── ans key
    ├── Key (Set A and B)                  # Answer key to verify
└── input-images/              # Sample OMR sheets and answer keys
    ├── Img1.jpeg
    └── Img2.jpeg
    ├── Img3.jpeg
    └── Img4.jpeg
    ├── Img5.jpeg
    └── Img6.jpeg
    ├── Img7.jpeg
    └── Img8.jpeg
    ├── Img9.jpeg
    └── Img10.jpeg
    ├── Img11.jpeg
    └── Img12.jpeg
    ├── Img13.jpeg
    └── Img14.jpeg
    ├── Img15.jpeg
    └── Img16.jpeg

🔧 Configuration

The system is pre-configured for Innomatics OMR sheet format:

• —5 Subjects: Python, EDA, SQL, Power BI, Statistics
• —20 Questions per Subject (Total: 100 questions)
• —4 Answer Choices: A, B, C, D per question

To modify for different formats, update the parameters in extract_bubbles() function.

Main Interface

• —Upload answer key and OMR sheets
• —Real-time processing with progress indicators
• —Individual sheet results display

Results Dashboard

• —Subject-wise score breakdown
• —Overall statistics and analytics
• —CSV export functionality

🐛 Troubleshooting

Common Issues:

1. 1.All subjects showing 0 scores:
2. 2.Ensure OMR sheet image is clear and well-lit
3. 3.Check that bubbles are properly filled (dark marks)
4. 4.Verify answer key format matches expected structure

1. 1.All subjects showing maximum scores:
2. 2.Enable debug mode to see detection details
3. 3.Check if image preprocessing is working correctly
4. 4.Verify answer key mapping is correct

1. 1."Too many values to unpack" error:
2. 2.This has been fixed in the latest version
3. 3.Ensure you're using the updated code files

Tips for Better Accuracy:

• —📷 Use good lighting when photographing OMR sheets
• —📐 Keep camera parallel to the sheet surface
• —🎯 Ensure all bubbles are clearly visible
• —🚫 Avoid shadows on the answer sheet

🚀 Deployment

This application is containerized using Docker for reliable deployment across different environments. The Docker configuration handles all system dependencies including OpenCV requirements.

👥 Author

• —Yaazhini S - Initial work - https://github.com/Yaazhini25/omr-evaluation-system