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
CNN BASED FEATURE EXTRACTION FOR AIR POLLUTION DETECTION
Authors Name:
R. Udaya Shanmuga
, Dr.G.Tamilpavai
Unique Id:
IJSDR2209142
Published In:
Volume 7 Issue 9, September-2022
Abstract:
The introduction of new techniques and methods for rapidly and reliably detecting and analyzing air quality is very essential and demanding for air pollution deduction. Machine learning algorithms equipped with basic features leads to poor and slow performance due to minimal complex image characteristics representation. Now, the deep learning methodology is used as the active and effective AI tool for feature extraction. Extracting complex image features from the given images is the standard and important procedure of feature extraction which utilizes the extracted features to predict the air pollution from the given image dataset. For this purpose convolution neural network is employed in this work. Air Pollution Image Dataset (APID) was created using publicly available camera images. Fully connected layers of CNN divides the features into different categories of different classes based on similarity. The first step is to number the images into 61 groups, and the second step was to group them again into seven groups. The best results were obtained using 4096 features with an accuracy of 67 percent and 95 percent, respectively, for 61 and 7 class groups. This provides an enhanced version of feature extraction and accurate results when compared with existing methods.
"CNN BASED FEATURE EXTRACTION FOR AIR POLLUTION DETECTION", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.7, Issue 9, page no.891 - 896, September-2022, Available :http://www.ijsdr.org/papers/IJSDR2209142.pdf
Downloads:
000336256
Publication Details:
Published Paper ID: IJSDR2209142
Registration ID:201821
Published In: Volume 7 Issue 9, September-2022
DOI (Digital Object Identifier):
Page No: 891 - 896
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn