INTERNATIONAL JOURNAL OF SCIENTIFIC DEVELOPMENT AND RESEARCH International Peer Reviewed & Refereed Journals, Open Access Journal ISSN Approved Journal No: 2455-2631 | Impact factor: 8.15 | ESTD Year: 2016
open access , Peer-reviewed, and Refereed Journals, Impact factor 8.15
Gopal Reddy K A
, Hamsa M N , Harish B , Rakshitha V
Unique Id:
IJSDR2304348
Published In:
Volume 8 Issue 4, April-2023
Abstract:
This According to the World Health Organization, around 40 million people are blind, and another 250 million have some sort of vision impairment. Unfortunately, those with vision impairments typically cannot view images. Our research proposes an approach that can automatically generate audio description of images, which can substantially assist the visually impaired. There are less opportunities for visually impaired people to interact in today’s environment, this paper is necessary. For those who are blind or visually impaired, using an image describer can be an enabler. The processed visuals that the visually handicapped cannot see are then given a suitable description and outputted as voice. A simplistic web-interface is developed using Streamlit . The paper uses the Inceptionv3 model as the feature extractor. The Transformer encoder-decoder model to extract image characteristics and generate the text description of the image, which is then converted to an audio using Google Text-to-Speech converter. The generated captions must now precisely reflect the image’s graphical information and be highly syntactically understandable. BLEU, METEOR, ROUGE-L, CIDEr score is used for evaluation after captioning. The results were shown to be more accurate, and as a result, they may help the blind access digital media.
Keywords:
Text description, Text-to-speech converter, Inceptionv3, Transformer encoder-decoder model, BLEU
Cite Article:
"Blind assistant tool using deep learning ", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.8, Issue 4, page no.2236 - 2241, April-2023, Available :http://www.ijsdr.org/papers/IJSDR2304348.pdf
Downloads:
000337072
Publication Details:
Published Paper ID: IJSDR2304348
Registration ID:204804
Published In: Volume 8 Issue 4, April-2023
DOI (Digital Object Identifier):
Page No: 2236 - 2241
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn