Smart Glasses for Visually Impaired Persons

Project Title

Smart Glasses for Visually Impaired Persons

Project Area of Specialization Artificial IntelligenceProject Summary 1. Introduction

 According to statistical analysis study of WHO (World Health Organization), approximately 285 million people around the world are blind or have amblyopia, 246 million of whom have serious vision problems. The increasing number of people with disabilities in world attracts the concern of researchers to invent various technologies, hoping that these technologies can assist the disabled people in carrying out their tasks in everyday life like normal people. So, we want to make something for them that would help them become independent. The conventional solutions to these situations are often seen to be the usage of guide canes to detect obstacles in front of them or relay on vocal guessing for identification of persons. Without eyes it is hard to survive even in his/her own environment, the person without sight has to memories everything in his environment and get irritated if someone misplaces objects. Navigation and or searching objects in an unfamiliar environment are surprisingly difficult for visually impaired people.

1.1 Overviews

The blind people normally travel using blind navigation system or by their accumulated memories in their long-term exploration. The main objective of the present work is to develop a low cost, reliable, portable, user friendly, low power and robust solution for smooth navigation. This project (Smart Glasses for Blind People), as meant are the glasses are for visually impaired people. It has an in-built sensor in it which spreads ultrasonic waves in the direction the person is going by scanning at most 1.5 meter. As soon as the obstacle is detected, the sensor detects it and sends it to the device which generates an automated voice in the earphone connected to the person's ear.[1]

These “Smart Glasses” are designed to help the blind people to recognize their relatives, whose ever image is already saved in the storage (Memory card or USB). The person will also be able to read and translate the typed text which is written in the English language. These kinds of inventions consider a solution to motivate blind students to complete their education despite all their difficulties. The glasses used many technologies to perform its tasks which are OCR.[2] The glasses are provided by Ultrasonic sensor which is used to measure the required distance between the user and the object. The maximum distance will be almost 1.5 meters.  All the computing and processing operations were done using the Raspberry Pi 4.

Project Objectives 1.5 Aims and Objectives

Following are the objectives.

• The blind navigation system is to guide blind and visually impaired people
• To help Blind people to walk without a walking stick.
• To reduce the dependency of the user on the other people.
• To identify the currencies.
• To identify known members or family members by name.
• To know the current location of the blind person.
• To send the location message to the specific person in case of need.
• To read the paper and hear by audio.

Project Implementation Method

The proposed solution is to overcome the limitation of the existing systems. To overcome these limitations, the proposed solution is based on deep learning algorithms plus compatibility and user-friendly hardware for practical implementation. SSD (SingleShot MultiBox Detector) is a popular algorithm in object detection. It's generally faster than Faster RCNN. It has no delegated region proposal network and predicts the boundary boxes and the classes directly from feature maps in one single pass. The solution is based on object recognition by which known persons and currencies identifications will be build. GPS and GSM is used for location and internet connection purpose. Algorithms is used for OCR (for reading) and to output in audio form.[4]

The technical area in the above Figure 1.1 is consisting of tiny credit card size computer raspberry-pi, pi-camera, GPS and ultrasonic sensor. When a blind person walking in any environment with the device in hand the obstacles that come in its way like a chair, bicycle and person etc, and when a known members or family members comes in front of blind so it will tell to blind his/her name in voice output. It is being detected through pi-camera with the help of Single Shot MultiBox Detector method for detecting objects in real-time images using a single deep neural network with the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. Secondly, the Ultrasonic sensor is used to find the distance between the object and the blind user. Both of the above data is processed by an embedded control unit raspberry-pi with the output in the voice to a blind. The output of the process will be the Name of the obstacle for example chair and secondly the location of the object for example bottom left the chair and finally the distance from the object like 1.5 meters.  The blind can also know the current location, GPS system of our smart glasses will tell the current location and also blind person can send location to care taker persons in case of need.

Benefits of the Project

For the growing competitive world around, it is quite difficult for a visually impaired person to move around independently and identify surrounding objectives correctly with ease. With the advancement of technology, there are several solutions but most of them have demerits such as low acceptance, high cost, difficult to usage etc. Based on the demand, devices supporting the visually impaired people has been introduced a long time now. On the other hand, keeping with pace, the more advanced algorithms and processing devices are introduced and progressing to higher accuracy and efficiency. This has inspired us to combine the concepts of implementing a processing device for serving blind individuals with a higher efficient methodology.

• The blind navigation system is to guide blind and visually impaired people
• To help Blind people to walk without a walking stick.
• To reduce the dependency of the user on the other people.
• To identify the currencies.
• To identify known members or family members by name.
• To know the current location of the blind person.
• To send the location message to the specific person in case of need.
• To read the paper and hear by audio.

Technical Details of Final Deliverable  Tools and Techniques

Here is the list of tools and equipment used to develop the proposed system.

• Microcontroller: Raspberry-pi-4
• Sensor: Ultrasonic Sensor
• Camera: Pi-camera
• Glasses
• Headphones
• SD card
• GPS
• Operating System: Raspbian
• graphical desktop-sharing system: Virtual Network Computing (VNC Viewer)
• Programming Language: Python.
• Program a Raspberry Pi:Terminal and Python IDLE. Program a Raspberry Pi with Python.

The proposed solution is to overcome the limitation of the existing systems. To overcome these limitations, the proposed solution is based on deep learning algorithms plus compatibility and user-friendly hardware for practical implementation. SSD (SingleShot MultiBox Detector) is a popular algorithm in object detection. It's generally faster than Faster RCNN. It has no delegated region proposal network and predicts the boundary boxes and the classes directly from feature maps in one single pass. The solution is based on object recognition by which known persons and currencies identifications will be build. GPS and GSM is used for location and internet connection purpose. Algorithms is used for OCR (for reading) and to output in audio form.

The technical area consisting of tiny credit card size computer raspberry-pi, pi-camera, GPS and ultrasonic sensor. When a blind person walking in any environment with the device in hand the obstacles that come in its way like a chair, bicycle and person etc, and when a known members or family members comes in front of blind so it will tell to blind his/her name in voice output. It is being detected through pi-camera with the help of Single Shot MultiBox Detector method for detecting objects in real-time images using a single deep neural network with the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. Secondly, the Ultrasonic sensor is used to find the distance between the object and the blind user. Both of the above data is processed by an embedded control unit raspberry-pi with the output in the voice to a blind. The output of the process will be the Name of the obstacle for example chair and secondly the location of the object for example bottom left the chair and finally the distance from the object like 1.5 meters.  The blind can also know the current location, GPS system of our smart glasses will tell the current location and also blind person can send location to care taker persons in case of need.

Final Deliverable of the Project HW/SW integrated systemCore Industry HealthOther Industries IT Core Technology Artificial Intelligence(AI)Other Technologies Internet of Things (IoT)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	50200
Rapberry pi 4	Equipment	1	30000	30000
Gps	Equipment	1	5000	5000
Camera	Equipment	1	3700	3700
SD card	Equipment	1	500	500
GSM	Equipment	1	6000	6000
Ultrasonic sensor	Equipment	1	500	500
Headphones	Equipment	1	500	500
3d print + Glasses	Equipment	1	2000	2000
Stationory	Equipment	1	2000	2000