DESIGN AND IMPLEMENTATION OF AN INTELLIGENT OBJECT RECOGNITION APPLICATION FOR VISUALLY IMPAIRED INDIVIDUALS

Abstract

Visual impairment can significantly affect an individual's ability to identify objects, interpret surroundings, and perform everyday activities independently. Although conventional assistive technologies provide important support, advances in artificial intelligence and computer vision have created opportunities for developing software-based systems capable of interpreting visual information and communicating it in accessible forms. This study focuses on the design and development of iSight, an object-recognition application intended to assist visually impaired individuals in identifying common objects within their environment. The proposed system utilizes computer-vision techniques to capture and analyze images through a device camera, recognize selected objects, and communicate the resulting information through an accessible output mechanism. The research considers important factors including recognition accuracy, response time, usability, accessibility, environmental conditions, and the limitations of automated recognition. The study is motivated by the growing global need for assistive technology and the increasing availability of artificial intelligence technologies capable of performing image recognition and interpretation. Current evidence indicates that billions of people require assistive products globally, while access remains uneven, particularly in low- and middle-income settings. The expected outcome is a functional prototype demonstrating how computer vision can be integrated into an accessible application to provide useful information about surrounding objects. The study contributes to research in artificial intelligence, computer vision, human-computer interaction, and assistive technology by demonstrating a practical approach to transforming visual information into accessible feedback. Ultimately, iSight is intended to complement—not replace—existing mobility and accessibility tools by providing additional environmental information that may support greater independence and participation among visually impaired users. Keywords: Artificial Intelligence, Computer Vision, Object Recognition, Object Detection, Machine Learning, Deep Learning, Assistive Technology, Visual Impairment, Accessibility, iSight.  

CHAPTER ONE

INTRODUCTION

1.1 Background to the Study

Vision is one of the major channels through which people perceive and interpret their surroundings. Consequently, significant visual impairment can affect a person's ability to identify objects, read information, recognize people, navigate unfamiliar environments, and perform everyday activities independently. The World Health Organization (WHO, 2026) estimates that at least 2.2 billion people globally have near or distance vision impairment, with at least 1 billion cases either preventable or yet to be adequately addressed. The effects extend beyond difficulty seeing; vision impairment can influence education, employment, mobility, social participation, and overall quality of life. Technology has increasingly been applied to reduce some of the functional barriers experienced by people with visual impairments. Assistive technology refers broadly to products, systems, and services designed to maintain or improve an individual's functioning and independence. According to WHO, assistive technology can promote participation in education, employment, civic life, and other aspects of society (WHO, 2026). The growing availability of smartphones, cameras, cloud computing, artificial intelligence (AI), and machine-learning frameworks has created new opportunities for developing software-based assistive solutions that can be more portable and potentially less expensive than specialized hardware. One promising area is computer vision, a branch of artificial intelligence concerned with enabling computers to interpret information contained in images and videos. Modern computer vision systems can perform tasks such as image classification, object detection, object recognition, scene understanding, and text recognition. The United States Patent and Trademark Office describes image and video recognition as involving the detection, categorization, identification, and interpretation of visual patterns. These capabilities make computer vision particularly relevant to assistive applications for people with visual impairments. Object recognition is especially important because visually impaired individuals may need assistance identifying objects in their immediate environment. An application capable of recognizing a bottle, chair, cup, book, vehicle, person, door, or other common object could provide useful environmental information that would otherwise be obtained primarily through sight. Recent research on assistive systems for visually impaired people identifies object detection, text detection, and text-to-speech as important technological components for improving independence and day-to-day interaction with the environment. The proposed iSight application is therefore conceived as a computer-vision-based assistive solution that captures visual information through a camera, processes the image using an object-recognition model, identifies relevant objects, and communicates the resulting information to the user through an accessible output mechanism, such as audio. Instead of requiring the user to interpret a visual display, the application can transform visual information into an understandable auditory response. The significance of this approach is strengthened by recent developments in deep learning. Conventional image-processing techniques often depend heavily on manually engineered visual features. Deep-learning approaches, particularly convolutional neural networks and related computer-vision architectures, can learn useful representations directly from large image datasets. This has contributed to substantial improvements in automated image classification and object detection. Current assistive-technology research increasingly combines computer vision with image-to-text and text-to-speech technologies to deliver real-time information to people with visual impairments. However, the development of an effective assistive application involves more than simply integrating an object-recognition model. The system must also consider usability, response time, recognition accuracy, accessibility, environmental conditions, privacy, and the practical needs of its intended users. Research into indoor scene-understanding systems, for example, emphasizes the importance of user-centered development and identifies use cases such as object finding, scene description, color detection, obstacle avoidance, and text reading. The need for accessible and affordable assistive technology is particularly relevant in low- and middle-income countries. WHO reports that access to assistive products remains limited in many such settings, with affordability, availability, financing, awareness, and trained personnel among the barriers. The WHO-UNICEF Global Report on Assistive Technology further estimates that more than 2.5 billion people require at least one assistive product globally. Against this background, the development of iSight provides an opportunity to explore how contemporary artificial intelligence and computer-vision techniques can be incorporated into an accessible software application. The project focuses on using object recognition not merely as a technological demonstration but as a practical mechanism for converting visual information into meaningful feedback for users with visual impairments.

1.2 Statement of the Problem

Visually impaired individuals may encounter difficulties when identifying objects within their immediate surroundings. Although traditional assistive methods such as white canes, guide assistance, Braille, audio descriptions, and other accessibility tools provide important support, they do not necessarily provide direct information about the identity of every object encountered in an environment. For example, a visually impaired individual may be able to detect that an object is physically present but may not immediately know whether it is a chair, bottle, table, bag, book, vehicle, or another item. In unfamiliar environments, this limitation can make independent interaction with the surroundings more challenging. Existing technological solutions have demonstrated the potential of computer vision to address some of these challenges, but access, usability, cost, technical complexity, and dependence on specialized equipment can limit their practical adoption. Research on wearable assistive devices has shown considerable technological development aimed at improving navigation, environmental understanding, and independence among visually impaired users. Nevertheless, there remains a need for accessible applications that integrate object recognition into a simple and practical user experience. Another challenge is that visual environments are dynamic. Objects may appear under different lighting conditions, at different distances, from different angles, or partially obscured by other objects. Therefore, an object-recognition application must be evaluated under realistic conditions rather than assuming that recognition will always be perfect. The problem addressed by this study is therefore the need for a practical software-based solution capable of using computer vision to identify selected objects and communicate the recognition results in an accessible form to visually impaired individuals.

1.3 Aim of the Study

The main aim of this study is to design and develop an accessible object-recognition application that uses computer vision and artificial intelligence techniques to identify objects in the user's environment and provide understandable feedback to visually impaired individuals.

1.4 Objectives of the Study

The specific objectives are to:

1. examine the challenges visually impaired individuals face when identifying objects in their surroundings;
2. review existing computer vision and assistive-technology approaches for supporting visually impaired users;
3. design an accessible architecture for an object-recognition application;
4. integrate an appropriate object-recognition model into the proposed application;
5. develop a mechanism for capturing and processing images through a camera;
6. provide recognizable and understandable feedback to the user through an accessible output mechanism;
7. evaluate the application's recognition performance under selected environmental conditions; and
8. assess the potential of the application to support greater independence in everyday object identification.

1.5 Research Questions

The study will be guided by the following questions:

1. What difficulties do visually impaired individuals experience when identifying objects in their environment?
2. How can computer-vision technology be applied to assist visually impaired users with object identification?
3. What system architecture is appropriate for an accessible object-recognition application?
4. How accurately can the proposed application recognize selected objects?
5. How effectively can recognition results be converted into understandable feedback?
6. What factors can influence the performance and usability of the proposed application?

1.6 Significance of the Study

The study is significant because it explores the application of contemporary artificial intelligence to an important accessibility challenge. For visually impaired individuals, an effective object-recognition application could provide additional environmental information and potentially support greater independence during routine activities. Visually impaired users may benefit from receiving information about objects that are difficult or impossible for them to identify visually. Researchers and students may use the project as a foundation for further research in computer vision, machine learning, human-computer interaction, accessibility, and assistive technology. Software developers may gain insight into designing applications that combine artificial intelligence with accessibility-oriented interfaces. Educational institutions may benefit from the project as an example of how emerging technologies can be applied to solve real-world social problems. Assistive-technology developers and organizations may also find the research useful when considering affordable, software-oriented approaches to accessibility. This aligns with the broader international emphasis on improving access to assistive technology and promoting inclusion.

1.7 Scope of the Study

The study focuses on the design and implementation of an object-recognition application intended to assist visually impaired individuals. The proposed system will primarily focus on recognizing selected common objects captured through a camera and communicating recognition results through an accessible output interface. The project will concentrate on software-based computer vision and artificial intelligence. It will not attempt to replace comprehensive mobility aids, professional rehabilitation services, guide assistance, or medical treatment. Similarly, the system should not be interpreted as guaranteeing complete environmental awareness or perfect object recognition. The evaluation will focus on factors such as recognition accuracy, response time, usability, and the application's ability to communicate recognized objects effectively.

1.8 Limitations of the Study

The proposed system may experience limitations associated with image quality, lighting, camera positioning, object distance, occlusion, background complexity, processing resources, and the limitations of the selected recognition model. Recognition models are also dependent on the data used during training. If an object is poorly represented in the training dataset, the system may produce an incorrect or uncertain prediction. Consequently, the application should be viewed as an assistive tool rather than an infallible source of environmental information. Internet connectivity may also become a limitation if some system components depend on cloud-based processing. A locally processed or hybrid architecture could reduce this dependency, although local processing may impose additional requirements on device hardware.

.9 Definition of Key Terms

Artificial Intelligence (AI): The field of computing concerned with developing systems capable of performing tasks that normally require aspects of human intelligence, such as learning, classification, and decision-making. Computer Vision: A field of artificial intelligence concerned with enabling computers to acquire, process, interpret, and understand visual information. Object Detection: The computational process of identifying the presence and location of objects within an image or video. Object Recognition: The process of determining the identity or category of an object represented in visual data. Assistive Technology: Products, systems, and related services designed to maintain or improve an individual's functioning, independence, and participation. Visual Impairment: A condition involving reduced or absent ability to see clearly, which can affect an individual's ability to perform visual tasks. Accessibility: The design of technologies and environments so that people with different abilities can use and interact with them effectively. iSight: The proposed object-recognition application developed in this study to provide accessible information about recognized objects to visually impaired users.