Please use this identifier to cite or link to this item:
http://dspace.dtu.ac.in:8080/jspui/handle/repository/16160
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | DIXIT, ANKITA | - |
dc.date.accessioned | 2018-08-21T12:28:51Z | - |
dc.date.available | 2018-08-21T12:28:51Z | - |
dc.date.issued | 2016-07 | - |
dc.identifier.uri | http://dspace.dtu.ac.in:8080/jspui/handle/repository/16160 | - |
dc.description.abstract | Understanding the emotion of attendees in a group session is an important aspect of today‘s learning world. The facial expression of a human being conveys a lot of information about identity and emotional state of the person. Emotion recognition is an interesting and challenging problem which gives details about the identity and emotional state of the person. Many experiments have been carried out in the past and different techniques are being proposed over the years for emotion detection that includes Eigen face based emotion detection and recognition, feature based emotion detection and recognition, machine learning based emotion detection and recognition and so on. However with the increasing migration of the application to be Convolutional Neural Network (CNN) most of the enterprise business operates from the CNN distance, with the advantage of enormous processing capabilities and online processing as well as device independent Application Program Interface (API). CNN provides a platform for almost all enterprise businesses. Therefore Deep Learning Based emotion detection and recognition is extremely important in order to model any enterprise system that takes any decision based on user‘s emotion. In this work we have proposed a unique Deep Learning Based emotion detection system. Multi emotions are detected using Histogram of Oriented Gradients (HOG) detector and Deep Learning. A system should be independent of the gender and age which the past systems have largely failed to provide. At the same time, our system should provide an enterprise level API‘s to data analytics such that recognition emotion data can be used from a large application perspective. In this project, we intend to propose a unique emotion detection based system over a CNN to analyse the behaviour of participants during the process of the session. The system should be able to acquire multiple faces, extract their emotion and put it into a CNN-database; further a data analytic system is implemented at CNN to extract meaningful, statistical insight and aggregated behaviour of users during the session. Overall system is capable of giving an insight about whether users are happy, skeptical or unhappy during the session. | en_US |
dc.language.iso | en | en_US |
dc.relation.ispartofseries | TD-4093; | - |
dc.subject | MULTI-EMOTION DETECTION | en_US |
dc.subject | HOG DETECTOR | en_US |
dc.subject | DEEP LEARNING | en_US |
dc.subject | CNN | en_US |
dc.title | MULTI-EMOTION DETECTION USING HOG DETECTOR AND DEEP LEARNING | en_US |
dc.type | Thesis | en_US |
Appears in Collections: | M.E./M.Tech. Computer Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
Ankita_Dixit_Samsung_MTech_Thesis_Final.pdf | 2.26 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.