A Study on Face Expression Observation Systems

International Journal of Trend in Scientific Research and Development (IJTSRD)
Volume 6 Issue 5, July-August 2022 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470
@ IJTSRD | Unique Paper ID – IJTSRD50450 | Volume – 6 | Issue – 5 | July-August 2022 Page 237
A Study on Face Expression Observation Systems
Jyoti1
, Neeraj Chawaria2
, Ekta2
1
Student, MERI College of Engineering & Technology, Sampla, Rohtak, Haryana, India
2
Assistant Professor, MERI College of Engineering & Technology, Sampla, Rohtak, Haryana, India
ABSTRACT
Human expressions can convey a great deal of information. We
cannot learn every language in the world, but we can decipher the
majority of human expressions. The state of a user's conduct in
various settings and scenarios can be inferred from their facial
expressions. Through various human-computer interface and
programming approaches, facial expression can be digitized. Face
detection, feature extraction, and kind of expression determination
are all parts of the facial expression perception process. Verbal and
non-verbal forms of communication are both possible. Through their
emotions, people can communicate nonverbally We've given a broad
summary of the various facial expression perception processes in the
literature in this article.
KEYWORDS: Face Expressions, Neural Nets, Knowledge, Statistics
How to cite this paper: Jyoti | Neeraj
Chawaria | Ekta "A Study on Face
Expression Observation Systems"
Published in
International Journal
of Trend in
Scientific Research
and Development
(ijtsrd), ISSN: 2456-
6470, Volume-6 |
Issue-5, August
2022, pp.237-243, URL:
www.ijtsrd.com/papers/ijtsrd50450.pdf
Copyright © 2022 by author(s) and
International Journal of Trend in
Scientific Research and Development
Journal. This is an
Open Access article
distributed under the
terms of the Creative Commons
Attribution License (CC BY 4.0)
(http://creativecommons.org/licenses/by/4.0)
INTRODUCTION
It is the procedure of classifying human emotions,
generally via facial expressions and via oral
languages. It is accomplished through human’s
perception inevitably and through computation
techniques. According to research, generally 90% of
the human communication is non-verbal, however
technique has emerged and usual code is not so good
for grasping a person’s stresses and intents. Now a
day, emotion recognition is also known as affective
Computation and is becoming available to many
developers. The process of emotion recognition has
been shown in Fig. 1.1.
Fig 1: Emotions Classification
IJTSRD50450

International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
RELATED WORK
Face recognition compares a face in a photo or video with a list of known faces to identify it. It is true that faces
must be entered into the system in order to build the collection of distinctive facial traits [1-2]. After then, the
system separates a new picture into its essential components and analyzes them to the data kept in the repository.
The emotions are classified on the basis of features analysis [3-4]. A lot of work has also been accomplished in
the field of image enhancement i.e. retrieval of an original image form a destroyed image [5-6], counting of
peoples in an image [7], image smoothing [8] and image inpainting methods [9].
TECHNIQUES OF EMOTION PERCEPTION
Generally, the process of recognition of emotions comprises of the investigation of human sentiments in multiple
configurations like texts, audios, or videos [10]. Several emotion categories are identified via the combination of
data from facial expressions, body passage, gesticulations, and language [11]. The methodology backs the
appearance of the emotive or emotional expressions. The present methods of emotion recognition can be
categorized into three specific emotion categories: On the basis of knowledge, according to statistics, and mixed
techniques [12].
Knowledge Related Techniques
Knowledge related methods are also known as lexicon methods. It employs field information, their meaning and
linguistic characteristics of speech to identifyspecific emotion categories [13]. This technique generally employs
knowledge related sources throughout the emotion category method like WordNet, SenticNet [14] ConceptNet,
and EmotiNet [15]. Main benefit of this technique is the convenience and budget carried out through the huge
accessibility of similar knowledge related documents. Also, this technique also has a drawback, i.e. its
incapability to manage perception distinctions and difficult linguistic guidelines. These methods may be
primarily categorized into two classes: dictionary related and quantity related methods. Dictionary related
methods get view or emotion source verses in a dictionary and find out their alternative and opposite word to
develop the primary list of thoughts or sentiments [16]. Quantity related methods begins with a source list of
view or sentiment words, and increase the datasets through discovering attributes with situation definite features
in a big quantity [16]. Although Quantity related methods consider situations but their enactment diverge in
dissimilar areas as a word in one area may have another meaning in next area.
Statistical Techniques
Statistical techniques usually include the practice of dissimilar supervised machine learning (ML)procedures that
have a huge set of marked data. This data is served to the procedures for the method to study and forecast the
suitable emotion forms. Usually, this method includes few groups of information: the training and the testing
groups. The training group is employed to study the qualities of the data. The testing group is employed to
authenticate the enactment of the (ML) procedure [17]. The ML procedures normally offer added sensible
categorization accurateness as compared to other methods, however it to obtain efficient outcomes it requires an
adequately big training group. Support Vector Machines (SVM) is one a popular ML procedure . There are also
unsupervised ML techniques which are used for emotion recognition, e.g., Deep learning [18]. These procedures
involve various configurations of (ANN) like CNN, LSTM, and ELM. In the area of emotion recognition, the
deep learning methods are highly accepted because of to its attainment in associated applications. See fig 2 [19]
for deep learning based emotion recognition.
Fig 2: Deep learning Based Emotion Perception
Hybrid Methods
Hybrid methods used for recognition of emotions are basically a mixture of info related methods and
mathematical approaches. These use corresponding features from both knowledge based and statistical

procedures. Certain processes have employed a collection of knowledge focused verbal components and
statistical approaches comprises of semantic computation and iFeel[20]. The function of these knowledge related
sources in the employment of hybrid methods is extremely significant in the emotion organization procedure. As
hybrid methods acquire knowledge from the profits offered by knowledge related as well as from statistical
methods. The hybrid approaches achieve improved grouping enactment as contrary to employing knowledge
centered or statistical approaches individually. The main drawback of consuming hybrid techniques is the
computation intricacy through the arrangement procedure.
Neural Nets for Emotion Recognition
A Neural Net is a subdivision of deep Learning and is a kind of technology which has turned to be widely trendy
since last term of years. Additionally, due to its mysterious capability to attain advanced up-to-date accurateness
for numerous sorting jobs, these have a precarious advantage which is massively supportive in emotion
recognition and in feature engineering spontaneously.
In case of the Neural Net, scientist may input the data in the form of transcript or vernaculars. Here the data
moves across various layers in the net. Every layer changes the inputs to attempt and transform it into somewhat
beneficial and analytical in the paradigm. Fig. 3 exhibits the working of convolution neural network for feature
extraction [21].
Fig 3: CNN Based Emotion Recognition
ANALYSIS OF THE EMOTION ASSESSMENT SYSTEM
To examine how facial expressions affect the mining of face features, authors in [22] created the Point
Distribution Model (PDM) technique. The location points (x, y) of the categorised or annotated points from the
training data set are analysed by the PDM method. Utilizing 180 photographs from 15 volunteers, the suggested
method selects 12 pictures from each person after each person chooses six sentiments. The collected features
from facial photos are categorised and matched using the Action Parameters (AP) Classifier. Authors in [23]
categorize CK facial sentiment and live videos through the SVM method to identify the universally recognised
feelings that are (in the scenario of the essential feeling of human supplied during instruction.
In order to obtain aspects from facial photos, authors in [24] introduced three independent feature extraction
techniques. The methods utilised were tested using different facial expression recognition parameters. The SVM
was used to categorise the retrieved features (SVM). The Cohn-Kanade (CK) database, which was constructed
from 100 individuals with ages ranging from 18 to 30, was used. For these studies, 310 photos were chosen from
the CK database. Authors in [25] studied three separate strategies for extraction of feature from face sentiments
for expression identification on 6 generally characterised primary emotions. These methods were discussed
individually. These methods are Singular Value Decomposition, FFT, and DCT. The retrieved facial features are
classified using the (SVM) classifier. The JAFFE database is used to conduct the investigation. There are 219
photos in this database. DCT+SVM technique, FFT+SVM technique, and SVD+SVM method all attained
recognition rates of 95 percent and 93 percent, respectively.
Authors in [26] studied three separate strategies for extracting features from face gestures for emotion
classification on 6 generally characterized primary emotions, especially angry and calm. These techniques were

discussed individually. These methods are Singular Value Decomposition, FFT, and DCT. The retrieved facial
features are classified using the (SVM) classifier. The JAFFE database is used to conduct the investigation.
There are 219 photos in this database.
Using the Discrete Wavelet Transform (DWT) method, authors in [27] retrieved characteristics from facial
photographs. They used 460 images they had taken from a collection of videos for their research. The (KNN)
technique and the (LDA) method were employed as two different classifiers after the features were extracted
from the images. LDA seeks to identify the best hyper plane for each of the six groups of sentiments (neutral,
happy, surprise and sad).
As feature extraction techniques, authors in [28] proposed the (WPCA) and (PPCA). The suggested algorithms
are thought to be the two most crucial techniques for reducing dimension and characteristic mining from a
person's facial region. Separate applications of the suggested techniques were made on the Image Database
(CKACFEID), which contains 500 photographs from 100 different people. For their investigations, researchers
chose 400 photos from this database. The proposed algorithms are assessed using Support Vector Machine.
To mine and encode facial traits, authors in [29] presented PHOG and Local Phase Quantisation (LPQ)
approaches. First, the PHOG technique was used to extract the features, and then PHOG and LPQ techniques
were combined. In these trials, 289 photos from the GEMEP-FERA dataset are used. Three techniques were
used by the researchers: the PHOG methodology with an SVM classifier.
To obtain traits from facial photos and increase the rate of recognition, authors in [30] integrated the (DCT),
(WT), (GF), and (GD) approaches. This research uses 213 photos from the JAFFE database that represent seven
different emotions: melancholy, anxiety, wrath, amazement, pleasure, calm, and contempt. Only 126 of the 213
total photographs were chosen for this experiment.
Authors in [31] recommend using the principal component analysis (PCA) approach to mine characteristics from
facial images to recognize or sense the person's sentiments (disgust, anger, sadness, happiness) in his face
picture. Authors collected characteristics and used the Singular Value Decomposition (SVD) classifier to
categories these features. For their experiments, they employed 50 training photos and 31 test pictures from the
JAFFE collection and the real database.
Authors in [32] used a different method to extract characteristics from facial photos called patch-based Gabor
characteristics. This method is defined in terms of maintaining the facial region statistics while mining local
attributes. The suggested technique is based on an approach used on the JAFFE dataset, which includes over 200
gloomy images for seven moods (six fundamental and one neutral). This research also makes use of the Cohn-
Kanade (CK) database. The SVM was used by researchers to categorise the retrieved features.
To determine the face emotions from a face image, authors in [3] employed a variety of techniques. To mine the
traits from a human face, utilise the (GLCM) approach. The features that are extracted are very effective and
take up little processing time. The key emotions recognised in this case are Joyful, Shock, Displeasure, Calm,
and Sadness. SVM is employed to train the retrieved attributes using different kernels. Researchers used the
Emotion Recognition Database in their research and got a classification rate of nearly 90%.
Table 1 displays a concise aspect of preceding readings which studied emotion approximation from facial
pictures. This table lists the names of the scientists, their techniques, databases, the overall number of photos,
expressions, train images, testing images, and the correctness they were able to obtain in each investigation.
Table 1: Analysis of Emotion Recognition from facial pictures
Sr.
No
References Techniques Databases Images Expressions
Trained
Image
Test
Image
Precision
(%)
1.
Chung-Lin &
Yu-Ming [22]
PDM+AP
Real DB 181
6 91 91 85
2.
Philipp & Rana
[23]
SVM
(CK&CK+)
and live video
80 6 91 88
3.
Tommaso,
Caifeng,
Vincent
& Ralph [24]
HOG+SVM CK 311 6 91 93
LBP+SVM CK 311 6 91 93
LTD+SVM CK 311 6 91 92

4.
KHARAT
&DUDUL [25]
DCT+SVM JAFFE 220 7 91 95
FFT +SVM JAFFE 220 7 90 95
SVD +SVM JAFFE 220 7 78 93
5.
Caifeng,
Shaogang and
Peter. [26]
LBP+SVM
MMI 80 6 90 87
JAFFE 214 7 90 80
6.
Murugappan,
Nagarajan, &
Yaacob [27]
DWT+KNN
Real
Database
461 5 91 84
DWT+ KDA
Real
Database
461 5 91 76
7.
Zhiguo &
Xuehong [28]
WPCA+SVM CKACFEID 501 6 100 400 89
PPCA+SVM
CKACFEID 501 6 100 400 85
8.
Abhinav,
Akshay, Yogesh
&Tom [29]
PHOG+SVM GEMEP 290 5 155 134 67%
PHOG&LPQ + SVM GEMEP 290 5 155 134 73
PHOG&LPQ+
LMNN
GEMEP 290 5 156 135 74
9.
Sandeep,
Shubh, Yogesh
&Neeta [30]
(DCT+WT+GF+GT)
+ADB
JAFFE 214 7 126 87 94
10.
Mandeep&
Rajeev [31]
PCA+SVD
JAFFE 15 6 7 7 99.99
Real
Database
83 6 50 31 99.99
11.
Ligang & Dian
[32]
Patch-based
Gabor+SVM
JAFFE 214 7 143 70 92.93%
CK 328 6 247 80 94.48 %
12.
Punitha &
Geetha [33]
GLCM+SVM
Facial
expression
DB
90 %
Conclusion
In this article, various methods are used to evaluate an
expression identification system from facial
photographs. However, classifying data is a
challenging process as well, especially when some
data are almost identical. This makes emotion
recognition from face photos a challenging work
because evaluating data is still complex. We
discussed some review of the literature that relate to
the topic of emotion recognition in facial expressions.
REFERENCES
[1] Sanju, Kirti Bhatia, Rohini Sharma, An
analytical survey on face recognition systems,
International Journal of Industrial Electronics
and Electrical Engineering, Volume-6, Issue-3,
Mar. -2018, pp. 61-68.
[2] Sanju, K. Bhatia, Rohini Sharma, Pca and
Eigen Face Based Face Recognition Method,
Journal of Emerging Technologies and
Innovative Research, June 2018, Volume 5,
Issue 6, pp. 491-496.
[3] Ankit Jain, Kirti Bhatia, Rohini Sharma, Shalini
Bhadola, An Overview on Facial Expression
Perception Mechanisms, SSRG International
Journal of Computer Science and Engineering,
Volume 6 Issue 4 - April 2019, pp. 19-24.
[4] Ankit Jain, Kirti Bhatia, Rohini Sharma, Shalini
Bhadola, An emotion recognition framework
through local binary patterns, Journal of
Emerging Technologies and Innovative
Research, Vol -6, Issue-5, May 2019.
[5] Jyoti, Kirti Bhatia, Shalini Bhadola, Rohini
Sharma, An Analysis of Facsimile
Demosaicing Procedures, International journal
of Innovative Research in computer and
communication engineering, Vol-08, Issue-07,
July 2020.
[6] Deepak Dahiya, Kirti Bhatia, Rohini Sharma,
Shalini Bhadola, A Deep Overview on Image
Denoising Approaches, International Journal of
Innovative Research in Computer and

Communication Engineering, Volume 9, Issue
7, July 2021.
[7] Nakul Nalwa, Shalini Bhadola, Kirti Bhatia,
Rohini Sharma, A Detailed Study on People
Tracking Methodologies in Different Scenarios,
International Journal of Innovative Research in
Computer and Communication Engineering,
Volume 10, Issue 6, June 2022.
[8] Mahesh Kumar Attri, Kirti Bhatia, Shalini
Bhadola, Rohini Sharma, An Image Sharpening
and Smoothing Approaches Analysis,
International Journal of Innovative Research in
Computer and Communication Engineering,
Volume 10, Issue 6, June 2022, pp 5573-5580.
[9] Yogita, Shalini Bhadola, Kirti Bhatia, Rohini
Sharma, A Deep Overview of Image Inpainting
Approaches, International Journal of Innovative
Research in Science, Engineering and
Technology, Volume 11, Issue 6, June 2022,
pp. 8744-8749.
[10] Poria, S., Cambria, E., Bajpai, R. and Hussain,
A. (2017). A review of affective computing:
From unimodal analysis to multimodal fusion.
Information Fusion, 37, 98–125.
[11] Caridakis, G. et al. (2007). Multimodal emotion
recognition from expressive faces, body
gestures and speech. IFIP the International
Federation for Information Processing, 247,
375–388.
[12] Cambria, E. (2016). Affective Computing and
Sentiment Analysis. IEEE Intelligent Systems,
31(2), 102–107.
[13] Rani, Meesala, S., and Sumathy, S. (2017).
Perspectives of the performance metrics in
lexicon and hybrid based approaches: a review.
International Journal of Engineering &
Technology, 6(4), 108.
[14] Cambria, E., Poria, S., Bajpai, R. and Schuller,
B. (2016). SenticNet 4: A Semantic Resource
for Sentiment Analysis Based on Conceptual
Primitives, In Proc. of COLING, the 26th
International Conference on Computational
Linguistics: Technical Papers, 2016, pp. 2666–
2677.
[15] Balahur, A., Hermida, Jesú, M. and Montoyo,
A. Detecting implicit expressions of emotion in
text: A comparative analysis. Decision Support
Systems, 53(4), 742–753.
[16] Madhoushi, Z., Hamdan, A., Razak, Z. and
Suhaila. (2015). Sentiment analysis techniques
in recent works, In Proc. of IEEE Conference
Publication, 2015, pp. 288–291.
[17] Sharef et al. (2016). Overview and Future
Opportunities of Sentiment Analysis
Approaches for Big Data. Journal of Computer
Science, 12 (3), 153–168.
[18] Sun, S., Luo, C. and Chen, J. (2017). A review
of natural language processing techniques for
opinion mining systems. Information Fusion,
36, 10–25.
[19] Xiaofeng Lu, Deep Learning Based Emotion
Recognition and Visualization of Figural
Representation, Front. Psychol., 06 January
2022, Human-Media Interaction.
[20] Araújo, M. et al. (2014). iFeel: a system that
compares and combines sentiment analysis
method, In Proc. of WWW '14 Companion.
ACM, 2014, pp. 75–78.
[21] Shiliang Zhang et al., Multimodal Deep
Convolutional Neural Network for Audio-
Visual Emotion Recognition, Proceedings of
the 2016 ACM on International Conference on
Multimedia Retrieval, 2016.
[22] Huang C. -L. and Huang, Y. -M. (1997). Facial
expression recognition using model-based
feature extraction and action parameters
classification. Journal of Visual
Communication and Image Representation, 8
(3), 278-290.
[23] Michel, P. and Kaliouby, R. E. (2005). Facial
expression recognition using support vector
machines, In Proc. of10th International
Conference on Human-Computer Interaction,
Crete, Greece, pp. 6-10.
[24] Gritti, T. et al. (2008). Local features based
facial expression recognition with face
registration errors, In Proc. of Automatic Face
& Gesture Recognition, 2008. FG'08. 8th IEEE
International Conference on, pp. 8-14.
[25] Kharat, G. and Dudul, S. (2008). Human
emotion recognition system using optimally
designed SVM with different facial feature
extraction techniques. WSEAS Transaction
Computing, 7 (6), 650-659.
[26] Shan, C., Gong, S. and McOwan, P. W. (2009).
Facial expression recognition based on local
binary patterns: A comprehensive study. Image
and Vision Computing, 27 (6), 803-816.
[27] Murugappan, M., Ramachandran, N. and
Sazali, Y. (2010). Classification of human

emotion from EEG using discrete wavelet
transform. Journal of Biomedical Science and
Engineering, 3(4), 390.
[28] Niu, Z. and Qiu, X. (2010). Facial expression
recognition based on weighted principal
component analysis and support vector
machines, In Proc. 2010 3rd International
Conference on Advanced Computer Theory and
Engineering (ICACTE), 2010, pp. 2-8.
[29] Dhall, A. et al. (2011). Emotion recognition
using PHOG and LPQ features, In Proc. of
Automatic Face & Gesture Recognition and
Workshops (FG 2011), IEEE International
Conference on. 2011, pp. 2-9.
[30] Gupta, S. K. et al. (2011). A hybrid method of
feature extraction for facial expression
recognition”, in Proc. of Signal-Image
Technology and Internet-Based Systems
(SITIS), Seventh International Conference
2011, pp. 47-52.
[31] Kaur, M. and Vashisht, R. (2011). Comparative
study of facial expression recognition
techniques. International Journal of Computer
Applications, 13(1), 43-50.
[32] Zhang, L. and Tjondronegoro, D. (2011). Facial
expression recognition using facial movement
features. IEEE Transactions on Affective
Computing, 2(4), 219-229.
[33] Punitha, A. and Geetha, M. K. (2013). Texture
based emotion recognition from facial
expressions using Support Vector Machine.
International journal of computer applications,
80 (5), 112-119.

A Study on Face Expression Observation Systems

Recommended

Recommended

More Related Content

Similar to A Study on Face Expression Observation Systems

Similar to A Study on Face Expression Observation Systems (20)

More from ijtsrd

More from ijtsrd (20)

Recently uploaded

Recently uploaded (20)

A Study on Face Expression Observation Systems