Smart Glasses for Visually Impaired
Our Final Year Project introduces Smart Glasses designed to assist visually impaired individuals by offering real-time support for reading and environmental awareness. The glasses use Optical Character Recognition (OCR) to read printed text aloud, combined with scene detection and object recognition to describe surroundings and identify everyday objects through audio output. Powered by a Raspberry Pi 4, the system includes a lightweight camera, headsets for quality feedback, image processing modules, and speech synthesis to deliver accurate and timely auditory information. This assistive solution is developed to enhance quality experiences and promote greater independence for the visually impaired.
- Hifsa Jahangir (F21605029) - Leader
- Rida Nadeem (F21605032)
- Maleeka Khan (F21605041)
- Momina Nauman (F21605047)
This project introduces Smart Glasses for Visually Impaired Individuals, addressing the challenge of accessing printed and digital text-an issue that significantly limits the social, professional, and educational opportunities of visually impaired people, particularly in countries like Pakistan. Existing assistive technologies, such as magnifiers, screen readers, and Braille systems, come with various limitations. Braille materials are costly and not universally available, while screen readers are limited to digital content and cannot interpret printed text in real time. Advanced solutions like OrCam MyEye and VisionCam offer real-time text-to-speech functionality, but they are often expensive, require internet connectivity, or lack support for local languages.
To address these gaps, we developed a cost-effective, offline, and user-friendly Smart Glasses system with real-time text-to-speech capabilities. The system uses a Raspberry Pi and a camera mounted on a wearable frame. It captures printed text from books, documents, and signage using Pytesseract (OCR) and converts it into speech using the offline pyttsx3 Text-to-Speech (TTS) engine.
Our prototype achieves approximately 90% accuracy in recognizing and vocalizing printed text. It functions effectively in diverse lighting conditions and does not require internet access, making it suitable for both urban and rural environments. Designed with affordability in mind, the system serves as a viable alternative to commercial devices, especially for users from low-income backgrounds.
In addition to enabling independent reading, the smart glasses have significant potential for applications in education and daily life. Future enhancements will include voice command features, scene description, object and currency recognition, and multilingual support, further increasing accessibility and utility.
This project offers a practical and inclusive solution to improve the quality of life for blind and visually impaired individuals by combining affordability, portability, and real-time functionality.
1. Accurate real-time audio feedback for printed text, objects, and scenes.
2. Enhanced independence for visually impaired users by providing clear and immediate audio descriptions of printed text and surrounding objects.
3. A wearable device that provides real-time verbal descriptions to assist users in everyday situations.
4. Reliable performance through the use of online tools and services.
5. A practical and affordable assistive technology designed to improve accessibility in educational and public environments.
Hardware Components:
o Raspberry Pi 4 Model B
o Pi Camera Module
o Audio output device (bone conduction headset or regular earphones)
o Power bank (for portability)
Software Stack:
o OCR: Pytesseract for extracting text from images
o TTS: pyttsx3 for offline text-to-speech audio output
o Object Detection: YOLOv5s model integrated with OpenCV for identifying objects in the environment
o Programming Language: Python
o Platform: Raspberry Pi OS
Workflow:
1. Image Capture
The smart glasses use a webcam to capture a live image of the user's surroundings.
2. Analysis Phase
The system processes the captured image to detect:
o Printed Text in the image
o Objects like a bottle, chair, or laptop
o Scene to understand the general environment (e.g., classroom, kitchen, computer lab)
3. Processing and Interpretation
o The system extracts any readable text from the image.
o It then identifies the environment or scenario in which the image was taken.
o Next, it detects and lists the objects present in that scene.
4. Audio Output Sequence
The information is read aloud to the user in a logical order:
o First, the detected text is spoken.
o Second, the scene or environment is described.
o Third, the identified objects are mentioned.
5. User Interaction
All responses are delivered through a connected headset or speaker, ensuring real-time audio assistance for visually impaired individuals.
