International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 05 | May 2024
p-ISSN: 2395-0072
www.irjet.net
Visual Assist:ML Based Object Detection For the Partially Impaired Rahul Thanvi1, Gaurav Mishra2, Yogesh Govari3 ,Vrushal Bagwe4, Prof. Nidhi Chitalia5 1BE student, Department of Information Technology St. Francis Institute of Technology Mumbai, India 2BE student, Department of Information Technology St. Francis Institute of Technology Mumbai, India 3BE student, Department of Information Technology St. Francis Institute of Technology Mumbai, India 4BE student, Department of Information Technology St. Francis Institute of Technology Mumbai, India 5Professor, Department of Information Technology St. Francis Institute of Technology Mumbai, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - In the face of evolving digital landscapes, the
entertainment, and notably, accessibility. This study explores the profound significance of accessibility within this evolving landscape, representing the combination of computer vision, machine learning, and artificial intelligence to create solution that enable people with disabilities to feel empowered. These solutions aim to close the divide between limitations and possibilities, particularly focusing on the challenges faced in the domain of accessibility.
'Visual Assist' project addresses the critical need for inclusivity through innovative usecases of machine learning and computer vision. This system empowers individuals with visual impairments by combining the advanced object detection capabilities of the YOLOv5 algorithm and the userfriendly audio generation of Google Text-to-Speech (gTTS). Recognizing the limitations imposed by visual impairments, Visual Assist aims to connect the information gap and foster independent engagement with the environment. Leveraging the real-time processing power and accuracy of YOLOv5, Visual Assist analyzes visual input from cameras, swiftly identifies objects within the user's surroundings, and conveys this information through clear and concise auditory descriptions generated by gTTS. This seamless translation of visual data into readily-understandable audio empowers users to perceive and navigate their surroundings independently and confidently.
Within the realm of accessibility, the project identifies major challenges, including the digital divide influenced by socioeconomic factors, standardization and interoperability issues in assistive technologies, ethical considerations surrounding privacy and bias in AI, and the performance optimization challenge for real-time applications such as Visual Assist. The rationale behind this research endeavor stems from a deep commitment to improving the lives of individuals with visual impairments in a technologically dominated world. Embracing the principle that technology ought be a bridge instead being a barrier, the project utilizes advanced machine learning and computer vision, specifically the YOLOv5 object detection algorithm and gTTS library, to provide audio feedback for objects in the environment. The overarching goal is to redefine accessibility, positioning technology as a method that get past barrier and offers new opportunities, guided by a vision of empowerment and inclusivity for all individuals, regardless of visual limitations.
Deployable across various platforms, including smartphones, Visual Assist grants individuals with visual impairments greater autonomy in navigating not only public environments but also their personal environments. Real-world testing and user feedback play a essential role in refining the system to ensure it effectively meets the needs of users. The potential impact of Visual Assist extends beyond enhanced mobility; it cultivates a feeling of self-assurance and independence for individuals with partial visual impairments, empowering them to confidently venture into their surroundings and discover new experiences.
2. LITERATURE REVIEW
Key Words: Visual impairment; Assistive technology; Object Detection; Computer vision; Machine learning; Accessibility; Inclusivity
The described AI powered visual assistance and combined reading support present a novel system for individuals completely blind, utilizing the RP3 Model B+ for its costeffectiveness and compact design. The system incorporates a camera, obstacle avoidance sensors, and advanced image processing algorithms for object detection. Additionally, it features an image-to-text converter using ultrasonic sensors for distance measurement. While the technology offers real-time feedback through an eSpeak speech synthesizer, it has limitations, such as potential inaccuracies in crowded
1. INTRODUCTION In the swiftly changing technological environment, the convergence of computer vision and machine learning has become instrumental in addressing real-world challenges. These cutting-edge fields leverage sophisticated algorithms and state-of-the-art technologies, influencing diverse sectors such as self-driving cars, healthcare,
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 1035