Abstract:

Presently, the visually impaired use a simple stick, which is insufficient to help them perform their daily activities independently. To resolve this issue, the paper proposes a smart and effective solution in the form of an e-stick module integrated with voice controlled android application.  The hardware e-stick module uses Convolutional Neural Networks to recognize objects and Optical Character Recognition for reading textual data in the surrounding. The software module/android application provides automated calling and messaging, live location tracking as well as stick tracking facilities. The primary aim is to provide a simple and affordable solution for the visually impaired by keeping the stick structurally similar to the traditional stick used by them today, that is thin, lightweight and easy to handle. All these functionalities are being provided at a lower cost and using efficient Natural Language processing (NLP) features. Thus, the proposed method attempts to help the visually impaired to lead a normal life.

Abstract

The visually impaired need to confront several problems while performing their daily activities. Independent navigation becomes one of the major problems in their social life. This issue further deteriorates when they travel across new unknown environments. This paper provides an efficient solution for the visually impaired in the form of a hardware automated stick based on Google's Cloud Video Intelligence API. This system uses real time video processing to analyze the obstacles or objects coming in the path of the blind and provides feedback in the form of voice messages. Hence the system facilitates real time navigation in both indoor and outdoor environments. Also the cost effectiveness and the versatility of the system make it more adaptive and easy to use.

Abstract

An image is an artefact that depicts visual perception and helps to capture a moment in life. With the advancements in computer vision and machine learning, it is now possible to analyse the content of the images and use it for various applications. Using advanced machine learning models of convolutional neural networks it is possible to detect objects or faces within images. Further, Optical Character Recognition can be used for the text extraction and analysis. As a whole, these techniques would contribute in classification of various images based on their content. This powerful analysis and classification of images can be integrated with natural language processing, augmented reality and web search to be used in several applications like Document Classification, Visual Positioning System, the Google Lens, or even in the gaming industry.