DOSTI

What it does

DOSTI is a assistive technology designed for the visually impaired individuals to assist them in day to day chores. It uses cutting edge technology to act as eyes for the visually impaired and help them navigate challenges while doing chores in their homes.

Your inspiration

I firmly believe that disability should not be considered a hindrance to achieve one's goals. The visually impaired should not be restricted from doing any daily activity, just because they are unable to see. I want them to live a normal life and carry out their daily chores. To implement this, I have created a technological solution named DOSTI. The word 'DOSTI' translates to 'friendship', in hindi which is my native language. So, while being a thoughtful and innovative idea, it also has a deeper meaning to itself and a connection with the user. It acts as a guide and a helper but more importantly as a FRIEND for the visually impaired.

How it works

DOSTI is a phone application which assists the visually impaired individuals in daily chores. It mainly consists of the 3 following components: Face Detector, Temperature Sensor, Image Detector I have implemented the following machine learning algorithms in our project: Speech Recognition (uses Natural Language Processing) , Text to Speech, Object detection, Colour detection, Position detection Face Detector: This component uses an OpenCV program. I have used the Haar Cascade Classifier for my prototype to be able to detect faces correctly. The current prototype is able to detect faces. Using Machine Learning and Artificial Intelligence, it can also be trained to recognise the names of the persons by their faces. Temperature Sensor: This component uses Arduino and LM35 Temperature Sensor to detect the temperature of nearby surroundings. Image Detector: Currently uses fine-tuned YOLO and certain online trained models for image and object detection.

Design process

Initially started as a standalone mobile application using a face detection program and image detection algorithm. I used Arduino (a micro computer) with an LM35 temperature sensor and integrated it with with the application. I plan to improve further by compressing the components into a wearable device such as a cap or a locket so that it is hands-free and user friendly.

How it is different

The most unique selling point is the temperature detection feature, combined with a speech detection and image detection technology. My product is specifically designed to assist the blind people in daily chores, especially chores in the kitchen. Most products for the visually impaired are designed around navigation. DOSTI focusses on giving the user more independence through a custom object detection model trained using YOLO specially for home and kitchen appliances. It also integrates natural language processing for speech recognition. Moreover, it helps enhance the user experience through a face detection feature which may be trained to be face recognition through uploading pictures of your helpers and family members.

Future plans

In terms of the design and development: 1. I aim to integrate an indoor navigation feature using stereo depth mapping for real time distance and direction calculation 2. Currently the program only uses face detection as a feature. But i aim to train it using machine learning to recognise faces of known people. 3. Finally, I aim to be able to combine the components in one single device connected to the mobile application through bluetooth. In terms of business plans, the market for devices for visually impaired is valued at over 4.2 billion USD. I wish to start small and test the product in local markets before moving on to expand it.

Awards

Presented at: Innovation Fair held by the National Science Centre, Delhi, India on 25-26 march 2023 Automate and Sustain contest held by Ryan international school, Delhi in october 2021 Toy Tech contest held by Maxfort School, Delhi on 27 august 2022