Object Detection to Assist Visually Impaired People: A Deep Neural Network Adventure

The object detection is used in almost every realworld application such as autonomous traversal, visual system, face detection and even more. This paper aims at applying object detection technique to assist visually impaired people. It helps visually impaired people to know about the objects around them to enable them to walk free. A prototype has been implemented on a Raspberry PI 3 using OpenCV libraries, and satisfactory performance is achieved. In this paper, detailed review has been carried out on object detection using region – conventionaal neural network (RCNN) based learning systems for a real-world application. This paper explores the various process of detecting objects using various object detections methods and walks through detection including a deep neural network for SSD implemented using Caffe model.

Download Full-text

Machine Intelligence to Assist Visually Impaired People: A Big Data Deep Neural Network Adventure

10.2196/preprints.9848 ◽

2018 ◽

Author(s):

Fereshteh S. Bashiri ◽

Eric Larose ◽

Jonathan Badger ◽

Zeyun Yu ◽

Peggy Peissig ◽

...

Keyword(s):

Neural Network ◽

Big Data ◽

Visually Impaired ◽

Deep Neural Network ◽

Machine Intelligence ◽

Visually Impaired People ◽

Impaired People

Download Full-text

Object Detection and Recognition for Visually Impaired People

International Journal of Engineering and Advanced Technology - Regular Issue ◽

10.35940/ijeat.c5568.049420 ◽

2020 ◽

Vol 9 (4) ◽

pp. 401-405

Keyword(s):

Neural Network ◽

Visually Impaired ◽

Deep Neural Network ◽

Sufficient Information ◽

Blind People ◽

The Real ◽

Visually Impaired People ◽

Impaired People ◽

Familiar Environment ◽

Audio Output

Good vision is an expensive gift but now a day’s loss of vision is becoming common issue. Blind or visually impaired people does not have any conscious about the danger they are facing in their daily life. To help the blind people the visual world has to be transformed into the audio world with the potential to inform them about objects. Various challenges are faced by visually impaired patients even in the familiar environment. Visually impaired individuals are at drawback due to lack of sufficient information about their familiar environment. This project employs a Convolution Neural Network for recognition of pre-trained objects. This project employs in deep learning a deep Neural Network (DNN) for recognizing the object which is captured from the real world. The captured image is compared with some pre trained objects that is stored in dataset .The comparative of the object is based on the shape and size of an objects. In deep neural network, TensorFlow package using a model called Mobile Net SSD that is comparing the real time capture image with pre trained object based on shape, size of the object. If the image is matched with that trained object, it will display the name of the object. Then the name of the object is converted into audio output with the help of gTTS. This will helps to identify and detect what object is present in front of blind people and give output as audio.

Download Full-text