A Major Project Abstract

On
AI-DRIVEN CLASSIFICATION & PREDICTION OF BLOOD GROUPS
THROUGH IMAGE PROCESSIN

Submitted to CMR ENGINEERING COLLEGE

BACHELOR OF TECHNOLOGY
IN
COMPUTER SCIENCE AND ENGINEERING
Submitted By
[Link] (228R1A0531)
[Link] SRI (228R1A0559)
[Link] (238R5A0506)

Under the guidance of

[Link] Priya
Assistant Professor, Department of CSE

Department of Computer Science & Engineering

CMR ENGINEERING COLLEGE
UGC AUTONOMOUS
(Accredited by NBA Approved by AICTE,NEW DELHI, Affiliated to JNTU Hyderabad)
Kandlakoya, Medchal Road, Medchal Malkajgiri Dist. Hyderabad-501 401)

2025-2026
 ABSTRACT

The project titled AI-Driven Classification & Prediction of Blood Groups through Image
Processing presents an innovative solution for rapid and accurate blood group determination
using Deep Learning Neural Networks (DNN) enhanced with Firefly Optimization (FFO).
Traditional blood typing methods require specialized equipment and are time-consuming. In
contrast, the proposed system relies solely on blood sample images and employs various
image preprocessing techniques such as de-noising, resizing, normalization, and
segmentation to isolate the blood region effectively. The core of this system is a DNN
algorithm, known for its powerful image classification capabilities. To further boost
performance, the Firefly Algorithm—a nature-inspired swarm intelligence method—is
integrated to optimize critical training parameters like learning rate, regularization factor, and
epoch count. This hybrid approach ensures high classification accuracy while reducing
computational complexity. The system workflow includes user registration, dataset loading
and preprocessing, training using Firefly-optimized DNN on 80% of the dataset, and
evaluation on the remaining 20%. Experimental results demonstrate the system’s ability to
classify blood groups with over 98% accuracy, validated through metrics such as precision,
recall, F1-score, and confusion matrix analysis. This project provides a lightweight, efficient,
and scalable solution for blood group prediction and can significantly aid healthcare
diagnostics, especially in resource-constrained [Link] lives.

1. Introduction

Blood group determination plays a critical role in modern healthcare. Whether it is for blood
transfusion, organ transplantation, surgery, or emergency medical interventions, knowing a
patient’s blood group accurately and quickly is of utmost importance. Traditionally, blood
typing is performed using serological methods, where specific antibodies are introduced to
blood samples and the resulting agglutination patterns are observed to classify the blood
group. While these methods are reliable, they require laboratory infrastructure, specialized
equipment, reagents, and skilled technicians. In many rural or resource-constrained
environments, these requirements are difficult to fulfill, resulting in delayed diagnosis and
treatment.
[Link] SYSTEM

In conventional medical practice, blood group determination is typically carried out using
serological testing methods that involve mixing a blood sample with specific antibodies and
observing agglutination reactions. These methods, while reliable, are often time-consuming,
require skilled technicians, and depend on laboratory-grade equipment such as centrifuges,
test tubes, and microscopes. In emergency scenarios or rural areas with limited resources,
these requirements become a bottleneck, delaying diagnosis and treatment. Additionally,
existing computer-aided systems for blood group detection are not widely available or depend
heavily on complex and expensive imaging devices, making them unsuitable for large-scale
or low-resource deployment.

2.1 DRAWBACKS OF EXISTING SYSTEM

There are some serious drawbacks in existing system. They are:

 Traditional serological techniques require multiple steps and human intervention,
leading to delays, especially in emergency medical situations.
 Specialized laboratory equipment, reagents, and trained professionals are necessary,
making the process costly and inaccessible in resource-limited settings.

[Link] SYSTEM

The proposed system offers an AI-based approach that leverages Deep Learning Neural
Networks (DNN) for the classification and prediction of blood groups directly from blood
sample images. This system integrates Firefly Optimization (FFO) to fine-tune DNN
hyperparameters such as learning rate, regularization, and number of epochs to maximize
classification accuracy. To ensure high-quality input, image processing techniques like de-
noising, resizing, normalization, and segmentation are applied before training. The system is
designed with user-friendly modules including registration, login, dataset loading, model
training, and blood group prediction. It processes 80% of the dataset for training and uses the
remaining 20% for testing, achieving over 98% accuracy. Unlike traditional methods, this
system is fast, requires minimal hardware, and provides an efficient, scalable, and cost-
effective solution for real-time blood group identification.
 Figure 3: Architecture of Proposed System

3.1 ADVANTAGES OF PROPOSED SYSTEM

The following are the advantages of proposed system:

· Provides rapid and accurate blood group prediction using image processing and deep
learning.
· Eliminates the need for costly reagents, specialized equipment, and highly skilled
personnel.
· Achieves high accuracy with Firefly-optimized DNN, making the system lightweight,
scalable, and suitable for resource-constrained environments.

[Link]:

Admin:

· In this module, the administrator will have the right to add, update, or delete blood banks,
hospitals, and doctors.
· Admin can send bulk messages/notifications to all registered users.
· Admin manages the overall functioning of the system to ensure smooth operation.
Donor:

· In this module, the donor must register into the system and can update their profile details.
· Donors can check messages or notifications posted by the admin or guests.
· Donors may be contacted in emergency situations for blood requirements.
Guest:

· In this module, the guest can view hospital, doctor, and blood bank details without
registration.
· Guests can post emergency messages visible to donors and admins.
· Guests have limited access compared to registered users.

[Link] REQUIREMENTS

 Processor : Pentium based systems/I3/I5

 RAM : 4GB(Minimum)
 Hard Disk : 40 GB or above

[Link] REQUIREMENTS

 Operating system : - Windows

 Coding Language : python.

Internal Guide Major Project Coordinator HOD