Emotion Classification from Facial Images and Videos Using a Convolutional Neural Network release_pp5ozwilynaobn4gdtjq3i7w7u

Published in International Journal of Advanced Trends in Computer Science and Engineering by The World Academy of Research in Science and Engineering.

2022   Volume 11, p8-13

Abstract

As a result of its wide range of academic and commercial applications, emotion recognition seems to be a important subject in computer vision & artificial intelligence. The majority of the decisions we make in our life are influenced by emotions. In this technology advancement, researchers found that properly categorising of human emotions may be a major source of development for companies in digital marketing. And that is what we will indeed be focusing on reading emotions of human being from facial image and videos. In the world of artificial intelligence, this concept falls under the category of cognitive systems. Facial expressions are essential to take into account while researching human behaviour including psychological characteristics. In this work, we used deep learning algorithms that recognise basic seven emotions through facial expressions (FER) and videos: happy, surprise, disgust, anger, neutral, sadness, and fear (VER). Deep learning has the potential to improve human-machine communication interaction because of its ability to learn features will allow machines to develop perception. To classify the emotions from facial images using deep learning techniques, we created the Convolution Neural Network Model and trained it on fer2013, a database of pre-recorded images with various emotions. And for emotion recognition from videos, we segment the video into individual frames at 30 frames per second and repeat the process of facial images on each frame, then do sentiment analysis, and finally reframe the emotional analysis output video with all of the available emotional individual frames.
In application/xml+jats format

Archived Files and Locations

application/pdf  1.3 MB
file_zwkue6eo3nhhtp47yarbg6ydau
www.warse.org (publisher)
web.archive.org (webarchive)
Read Archived PDF
Preserved and Accessible
Type  article-journal
Stage   published
Date   2022-02-07
Journal Metadata
Not in DOAJ
Not in Keepers Registry
ISSN-L:  2278-3091
Work Entity
access all versions, variants, and formats of this works (eg, pre-prints)
Catalog Record
Revision: 591b4fdf-da27-4fa2-b503-4d00df9a4371
API URL: JSON