Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY1072

OMR Automated Grading

Janardhan Singh K.
Assistant Professor, Department of ISE, RNSIT, Bengaluru

Sanjay Kulkarni (RN20IS138)1; Sanket B Patil (1RN20IS139)2; Shashank M (1RN20IS145)3; Shashanka UN (1RN20IS147)4

Abstract:- The paper highlights the necessity for a aiming to democratize access to automated grading processes
technologically advanced system capable of efficiently while reducing costs. By capturing and processing images of
grading multiple-choice question (MCQ) exams through answer sheets using content filtration and image processing
webcam-based evaluation. MCQ-style assessments have algorithms, the system aims to revolutionize assessment
gained widespread use in educational and organizational methods, ensuring efficiency and accuracy without using
settings due to their effectiveness and time-saving expensive hardware or software. Additionally, the system's
advantages. However, manually grading these exams algorithm is made to find standardized answer sheet formats,
presents significant challenges. Managing a large number ensuring compatibility across various exam layouts and
of answer sheets in a timely manner is labor-intensive and educational settings.
error-prone, potentially leading to scoring discrepancies.
Additionally, the logistical burden of storing and handling Moreover, OMR grading technology is expected to
physical answer sheets is cumbersome, with risks such as alleviate the inefficiencies and inaccuracies of manual grading,
damage from environmental factors like fire or moisture. enhancing the educational assessment ecosystem. By
While larger institutions may utilize specialized Optical automating the recognition and interpretation of marked
Mark Recognition (OMR) technology for grading, smaller responses, educators can redirect their focus towards refining
educational entities often lack access to such costly instructional methodologies, engaging with students, and
equipment. To address these challenges, the paper proposes pursuing professional development opportunities. Additionally,
an innovative solution: leveraging webcam technology to the system's scalability ensures seamless handling of diverse
automate the grading process. By capturing images of question formats and extensive surveys or examinations,
answer sheets and employing sophisticated content-filtering catering to the evolving demands of educational institutions
and image processing algorithms facilitated by the OpenCV worldwide.
library, the system can accurately interpret and evaluate
marked answers. Overall, the proposed system represents a The scalability of the proposed system extends beyond
significant advancement in exam grading methodology, volume management to encompass quicker turnaround times in
providing a practical and cost-effective solution to the the grading cycle, facilitating timely feedback to students and
longstanding challenges associated with manual grading of supporting their academic growth by promptly addressing any
MCQ-based assessments. By integrating webcam misconceptions or knowledge gaps. Furthermore, the addition
technology into the grading process, the system aims to of OMR technology fosters a standardized and consistent
enhance efficiency and accuracy while catering to the needs grading approach, minimizing variations that may arise in
of various educational and organizational assessments. manual grading and promoting fairness and impartiality in
evaluations. The automation of response processing not only
I. INTRODUCTION optimizes workflow efficiencies but also contributes to the
establishment of a reliable and transparent evaluation system.
The introduction highlights the widespread use of
multiple-choice question (MCQ) exams worldwide and their Educators stand to gain significantly from embracing
significance in evaluating students' academic abilities. It OMR technology, benefiting from reduced errors and increased
emphasizes the prevalence of MCQ formats in prominent scalability. Overall, OMR technology signifies a notable
exams such as K-CET, NEET, and also in international advancement in educational assessment methods, transforming
assessments like TOEFL, GRE, and GMAT, owing to their grading practices and promoting a more effective, fair, and
time-saving nature and streamlined evaluation processes. student-centered learning environment. The paper stresses the
However, traditional grading methods facing challenges mainly importance of innovative solutions to overcome from the
in such institutions which have lack access to advanced problems associated with traditional grading methods and
scanning equipment, leading to time-consuming manual highlights the proposed system's capability to simplify grading
grading processes, particularly with large numbers of test- procedures while reducing expenses and resource demands.
takers. To address these challenges, the paper proposes a novel Through sophisticated content filtration and image processing
solution utilizing the built-in webcams of laptops for OMR, algorithms, the system accurately captures and assesses

IJISRT24MAY1072 www.ijisrt.com 3757

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY1072

responses from MCQ answer sheets, catering to diverse C. Identify Marked OMR Option:
institutional requirements worldwide. The software analyzes the scanned image to identify
marked responses made by the student. It detects darkened
The paper emphasizes the critical necessity for creative bubbles or checkboxes corresponding to selected options on the
solutions to address the challenges linked with conventional OMR sheet.
grading methodologies. Through the implementation of
sophisticated content filtration and image processing D. Storing the Identified Data:
algorithms, the system ensures the accurate capture and Store the recognized data, comprising student
assessment of responses from multiple-choice question (MCQ) identification particulars and chosen responses for each query.
answer sheets. This capability makes the system adaptable to The retention of this data enables analysis, reporting, and
diverse institutional requirements worldwide, catering to feedback generation centered on student performance.
various exam formats and educational contexts.
E. Validate the Chosen Option:
Compare the student's responses with correct answers
from the answer key. Validated responses contribute to the
student's score, while discrepancies are flagged for review. This
step is used to ensuring fairness and accuracy in grading, as it
helps identify and rectify any errors or inconsistencies in the
evaluation process.

F. Evaluate Score:
Based on validation, calculate the student's OMR test scores.
Correct answers add to final score, while incorrect responses
may result in deductions.

G. Storing the Validated OMR Sheet:

Store the validated OMR sheet, along with the score and
metadata, for future reference. Archived sheets can be accessed
for analysis, review, or record- keeping purposes.

II. ALGORITHM AND IMPLEMENTATION

 Median Filtering: Median filtering is a widely employed

method in image processing aimed at enhancing the quality
of images by reducing noise. It achieves this by calculating
pixel values based on the median value of neighboring
pixels in its vicinity. This technique efficiently reduces
sudden pixel value variations due to noise, leading to clearer
and aesthetically improved images.
 Conversion to Grayscale: RGB, an acronym for red, green,
Fig 1: Application Flow and blue, is a color representation system commonly used
in digital screens to create wide range of colors. Converting
A. Manually Entering Correct OMR Answers: to grayscale entails changing an image's RGB format to a
Prepare an answer key with the accurate responses before single-channel grayscale format, with each pixel
scanning the OMR sheets. The answer key serves as a represented by an intensity value spanning from black to
reference for comparing scanned responses during evaluation, white. This conversion removes color information, resulting
ensuring grading accuracy. in a monochromatic image that emphasizes luminance and
simplifies subsequent image processing tasks.
B. Start Webcam:  Edge Detection: Edge detection is a fundamental which
Activate the webcam hardware on the computer or device helps in identifying the boundaries or edges of objects in an
to initiate the scanning process. Webcam activation enables the image. By detecting abrupt changes in pixel intensity, edge
device to capture images of filled OMR sheets for subsequent detection algorithms highlight regions of significant
processing. Scan the Student OMR Sheets: Place the filled contrast, which typically correspond to object boundaries.
OMR sheet under the webcam's view for scanning. The This step plays a crucial role in a range of applications such
webcam captures an image of the sheet, which is processed by as object detection, image segmentation, and feature
the OMR software. extraction. It enables the identification and delineation of

IJISRT24MAY1072 www.ijisrt.com 3758

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY1072

objects or details within images, paving the way for deeper and is widely used in applications like image segmentation
analysis and interpretation. and object detection.
 Complementation of the Image: The complement of an  Canny Edge Detection Algorithm: The Canny edge
image alters its brightness by reversing the intensity values detection algorithm stands as a foundational tool in
of each pixel. Dark color portion of the image become computer vision and image processing, renowned for its
lighter, while light areas become darker. This conversion precision and effectiveness in edge detection. Conceived by
plays a vital role in adjusting the image's contrast and John F. Canny in 1986, the algorithm begins by converting
dynamic range, thereby improving its visual appeal and the image from RGB to grayscale, streamlining the edge
simplifying subsequent image processing tasks. Through detection process. It employs various methods such as
intensity value inversion, the complement operation reveals Gaussian smoothing to diminish noise, computes gradients
obscured details caused by extreme brightness or darkness to determine edge strength and orientation, and performs
in the original image, enhancing its clarity and overall non-maximum suppression to refine detected edges.
quality.
 Thresholding: Thresholding simplifies images by and hysteresis thresholding to link adjacent edge pixels.
converting them into binary form using a specified threshold By combining these techniques, the Canny algorithm can
value. Pixels above the threshold are categorized as one accurately detect and localize edges while minimizing false
type, while those below it are categorized differently. This detections and noise artifacts, making it a vital tool in various
method aids in focusing on specific details within the image image analysis and computer vision applications.

Fig 2: Flow Chart

 Scan OMR Sheet: The OMR sheet, filled out by the student,  Find Edges: Algorithms for edge detection are utilized on
is scanned using an optical scanner or imaging device. This the grayscale image to outline the boundaries of circles
process converts the physical sheet into a image file. marked on the OMR sheet, denoting the areas filled by the
 Grayscale Conversion: The scanned image is converted student.
from color to grayscale. Grayscale images represent pixel  Defining Boxes: Using the identified edges, bounding
intensities using shades of gray, simplifying subsequent boxes are established around each marked circle to isolate
image processing steps while retaining essential information and concentrate on these specific regions, aiding in further
for analysis. analysis.

IJISRT24MAY1072 www.ijisrt.com 3759

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY1072

 Thresholding: Employing a thresholding technique on the  Check Student Answered: The system checks whether the
grayscale image converts it into binary form, categorizing student has answered any questions on the sheet. If
pixels as black or white based on a defined threshold. This questions are marked, the system compares the marked
step segregates marked circles from background elements array with the original array to identify which questions
or shadows, enhancing clarity. have been answered.
 Split Rows: The binary image is segmented into distinct  Compare: The marked responses on the sheet undergo
rows, representing different sections or questions on the comparison with the corresponding answers from the
answer sheet. This segmentation organizes the data for answer key, establishing the correctness of each student-
focused processing, allowing examination of individual selected answer.
rows sequentially.  Calculate Score: Using the comparison outcomes, the
 Find Marked Circles: Each row in the binary image system calculates the student's score by tallying the number
undergoes analysis to identify marked circles, indicating the of correct answers and assigning points accordingly.
student's selected answers. This detection process involves  Display Score: The system displays the student's score,
pinpointing filled circles within the previously established providing feedback on their performance by showcasing the
bounding boxes. total correct answers and the resulting overall score. Lastly,
the student's answer sheet and score are saved as a file for
archival and future analysis. This storage enables ongoing
tracking and referencing of student progress and
performance trends.

III. RESULT

Fig 3: Result

 Original: This image represents the original input image  Canny: Canny edge detection is then applied to the blurred
loaded from a file. image, identifying edges as regions with significant
 Gray: After capturing the original image, it's converted to intensity changes, crucial for object outline identification.
grayscale. Grayscale images have only shades of gray,  Contours: The outlines of objects detected in the Canny
which simplifies the subsequent image processing steps. edge-detected image are represented as contours, visualized
 Blur: The grayscale image is then subjected to Gaussian on the original image.
blurring, which helps in reducing noise and unwanted  Biggest Contour: The largest contour, usually representing
details. the main object of interest, is identified among all detected
contours for next process.

IJISRT24MAY1072 www.ijisrt.com 3760

Volume 9, Issue 5, May – 2024 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.38124/ijisrt/IJISRT24MAY1072

 Wrapped: The largest contour, likely containing the main streamlined and equitable assessment process in educational and
rectangular region with multiple-choice questions, organizational settings.
undergoes a transformation to a rectangular shape for
alignment. V. FUTURE WORKS
 Threshold: The warped image is then threshold to segment
regions corresponding to filled bubbles in multiple-choice In future developments of the automated grading system,
questions, simplifying object identification. several enhancements are proposed to enhance its functionality and
 Result: The final result image displays the processed input usefulness for educators and students alike. Firstly, expanding the
along with detected answers and grading. system to support grading in multiple languages would
 Img Raw Drawing: This image likely shows intermediate accommodate diverse student populations, ensuring fair and
processing steps or debugging information, such as accurate grading for all regardless of their language proficiency.
drawings on a blank canvas. This addition would contribute to inclusivity and accessibility in
 Inv Wrap: The inverse warped image is the result of the assessment process.
reversing the perspective transformation for overlaying
detected answers and grading back onto the original input Secondly, implementing a feature for instant feedback
image. students on their performance would promote continuous
 Final: The final image combines the original input with learning and improvement. This feature could analyze students
detected answers and grading, providing a complete view of responses and provide personalized feedback, helping them
analysis results. identify strengths and areas for improvement promptly.

IV. CONCLUSION Additionally, offering cloud storage options for storing

evaluated answer sheets addresses concerns related to data
In various applications, such as academic assessments or security and accessibility. Cloud storage provides a secure and
performance evaluations, grades serve as a standardized centralized repository for storing graded answer sheets,
method for categorizing marks. For instance, an A grade ensuring easy access for educators while maintaining data
typically signifies a score above 90, while a B grade may integrity and confidentiality.
indicate a score above 80. This grading system provides clarity
and consistency in evaluating individuals' achievements or Lastly, implementing a feature that offers detailed scores
proficiency levels across different contexts. for different sections of the exam further enriches the grading
system's capabilities. By providing granular insights into
Leveraging the prowess of OpenCV, a powerful open- students' performance across various exam sections such as
source computer vision library, ensures precise image technical areas, coding, and aptitude, This detailed scoring
processing and grading. OpenCV's sophisticated algorithms system allows for targeted interventions and personalized
enable the system to accurately interpret and analyze scanned support, ultimately contributing to improved learning
answer sheets or images, thereby enhancing the reliability and outcomes.
accuracy of the grading process.
REFERENCES
Compared to traditional manual grading methods, which
often involve extensive human labor and time-intensive [1]. Fairley R. (2002) Software Engineering Concepts (For
processes, the proposed system offers a more cost-effective project size) New York: Tata Mac Graw Hill.
solution. By automating the grading process, institutions can [2]. A. M. Smith, “Optical mark reading - making it easy for
substantially cut costs tied to with hiring personnel and users”, In Proceedings of the 9th annual ACM
investing in specialized grading equipment, resulting in SIGUCCS conference on User services, United States,
substantial cost savings over time. Furthermore, the versatility 1981, pp:257-263.
of the system extends beyond exams, as it can also be employed [3]. Ngo Quoc Tao and Do Nang Toan “Some Charactistical
for tasks such as surveys and attendance tracking. Its aspects of Markreader Software Package for Automatic
adaptability to various assessment scenarios enhances its utility Mark Data Entry”, Circuits and Systems, 2002.
and value for educational institutions seeking efficient solutions APCCAS'02, pages 437 - 442 vol.2, 2002
for grading and data collection purposes. [4]. Palmer, Roger C., The Basics of Automatic
Identification [Electronic version] Canadian
The proposed system represents a significant leap forward Datasystems, 21 (9), p. 30-33, Sept 1989.
in automated grading technology, offering a comprehensive [5]. Jalote P. (2005) Software Project Management In
solution that prioritizes efficiency, accuracy, and accessibility. Practices-3rdEdition, United States of America:
By combining innovative algorithms, and cost-effective Springer Science + Business Media, Inc.
implementation, the system addresses longstanding challenges
in grading methodologies, paving the way for a more

IJISRT24MAY1072 www.ijisrt.com 3761