Zena Tarik Mohammed*
| Ielaf O. Abdul Majjed Dahl
© 2024 The authors. This article is published by IIETA and is licensed under the CC BY 4.0 license (http://creativecommons.org/licenses/by/4.0/).
OPEN ACCESS
Recently, the need to separate truth from lies has motivated lie detection as a constant human endeavor; therefore there is a need to develop lie detection techniques and focus on the new area of lie detection utilizing facial expression. Human faces are a powerful repository of emotions in the complicated interaction between verbal and non-verbal clues that characterize human communication. From this micro-expression, the transitory emotion discloses the more prominent indicators that precede deceitful behavior, which makes the tapestry rich in information that can be harnessed to detect a lie. Historically, the development of deceiving lies passed through many developments to find the best way to get high performance, but the development of artificial intelligence and face recognition has further altered the landscape of lie detection. In this paper, the reason for lie detection is revealed with the techniques used to detect lies. The paper aims to present and survey the techniques with comparison used to detect lies, which will highlight the importance of this topic and urge researchers to develop current techniques or find other related techniques that serve the issue. The presentation of the techniques in this research revealed that the lie detection technique using facial expressions is considered the best technique to achieve the detection of lies. Facial expression is the most efficient because it does not require physical contact and because they are visual of real internal feelings and not voluntary movements, and computer vision and artificial intelligence have had an effective role in supporting this method and exploiting it optimally. Finally, the paper shows the limitations and achievements that the researchers found in their research to help researchers in this field.
behavior analysis, facial expression, lie detection technique, micro-expression
The human can differentiate truth. Scholars, scientists, and investigators have been fascinated by the phenomenon of deception for ages, regardless of whether it is motivated by self-preservation, self-interest, or a variety of other factors. The desire to uncover the hidden and reveal the truth beneath a curtain of lies has sparked the examination of numerous techniques and tools designed to detect deception's subtle signals. The ability to detect lies in people's comments and behaviors has become one of these tools, and lie detection is a growing subject of study and practice. Over the years, we have discovered the importance of detecting lies in many places [1, 2]:
1. Criminal investigation and law court where it is important to present lie detection evidence using a lie detection method or expert testimony of facial expression [3, 4].
2. Border control, which is needed at border crossings and airports to identify those exhibiting questionable conduct or giving lie information [5-7].
3. Employment and workplace, where we use it in interviews and workplace investigations [8].
4. Therapeutic assessment and forensic psychology, where psychologists and therapists need to assess the emotional states and truthfulness of clients, aiding in the diagnosis and treatment of mental health, also use the lie detection method in situations including testimony from witnesses or assessments of people’s mental states [9].
Intelligence and counterintelligence are where we used to find spies, moles, or people with dual loyalties within their organization [10].
Consumer protection and fraud prevention are where financial institutions and consumer protection organizations can spot fraudulent behavior like identity theft or dishonest business practices [11].
Finally, use it in academic research and experimental studies where we use methods to study the behavior of humans and nonverbal communication and measure the impact of deception on human behavior [12-15].
This paper addresses the age-old question, "Can deception truly be unraveled, or will it forever remain an enigmatic facet of the human experience?" We explore various lie detection techniques, highlighting their pros and cons, and focus on facial expression analysis. The goal is to guide researchers in applying artificial intelligence methods to enhance and expand datasets in this area.
This makes it important to detect lies [2, 16, 17]. There are five techniques to detect lies, and each method has advantages and disadvantages Table 1 shows that each technique has its approach and principles for assessing deception or truthfulness.
Table 1. Lie detection techniques comparison
|
Techniques |
Working |
Accuracy |
Advantages |
Problems |
|
Polygraph method |
It identifies stress related reactions such as increase in skin conductivity, heart rate, breathing rate and BP (blood pressure). |
Studies in this field vary considerably, which shows 60 to more than 90 percent accuracy. |
Cost of the system has decreased due to competition in the market. |
Human can experience stress for many reasons: Tension, annoyance, pain, fear, surprise. 2. An expertise need to read the graphs |
|
Voice-stress analysis method |
This needs a computer program which determines variations in speech patterns, such as tone, pitch and intensity, stress, which may imply deception. |
Uncertain |
*Not required any direct contact with the subject. *Ease of use, only a microphone required to attached to the subject. |
*Found stress better, than a lie but if it is a good lie detector it remains unclear. |
|
Micro-expressions |
Popularly working on facial movements which can reveal range of emotions, including however a person is trying to hide |
Highly accurate |
*Not required any intervention with the subject. *Could be used secretively by videotaping *High precision |
Availability of datasets *Compound expressions |
|
Functional MRI |
fMRI signifies changes in the brain activity based on the concept that when people lie, they shows greater changes in blood flow rate. |
Unclear |
*FMRI has no risk, as it does not use radiations like X-Rays. *It can evaluate brain function safely. |
*It is expensive *fMRI scan difficult to interpret. *Researchers still don't completely understand how it works. |
|
Thermal imaging |
It measures the surface skin temperature accurately. |
Accurate for measuring surface skin temperature. |
*Works well without Physical interaction. *It can act as a cue to detect deceit. |
*High initial cost. *High temperature may be the reason of many other causes which may produce wrong results. *Environment sensitive |
2.1 Polygraph method (polygraphy)
Used to measure physiological responses such as heart rate and blood pressure while the device connects to the skin while the person answers the question, the advantages of this method are that it can be widely used and that real-time monitoring makes the investigators and jurors understand, while the challenges are that psychological stress may produce a false result and skilled people may influence the outcomes, making it difficult to detect lies [18] (Figure 1).
Figure 1. Classical polygraph method [19]
2.2 Voice stress analysis method (phonoscope)
Non-invasive and does not need to require physical control, the advantage of this method is that it can provide real-time results during speech, and it’s useful in interviews and interrogations. While the challenges are limited in scientific support, this means that many studies have failed to demonstrate consistent and accurate results using this method, and changes in vocal characteristics can result from factors such as nervousness or illness [20] (Figure 2).
Figure 2. Stress analysis method [21]
2.3 Brain imaging methods
Non-invasively examine the brain’s anatomy, physiology, and neural activity. It is used to visualize and study the structure, function, and activity of the brain. The advantages of this method are that it provides information on the neurological mechanisms involved in deception as well as unbiased, potentially less manipulable data on brain activity. The disadvantages are that this method is complex, costly, and typically not suitable for real-time or field applications. Also, it is limited due to the need for specialized equipment, controlled environments, and expert analysis, and finally, brain imaging raises ethical concerns related to privacy, consent, and the potential for misinterpretation of brain data [22, 23] (Figure 3).
Figure 3. A Change in the brain when lying [24]
2.4 Thermal imaging method
Non-invasive, it uses infrared technology for the detection and visualization of the heat patterns emitted by the human body. The advantage of this method is that it can record changes in skin temperature and blood flow and offer real-time feedback when a person answers questions or converses with others, while the disadvantage is that stress can lead to skin temperature and blood flow changes. These changes are not caused by lying; they may be caused by nervousness or discomfort during the process. Environmental factors may influence skin temperatures, such as room temperature and humidity, and various outside temperatures lead to various thermal imaging readings. Since there aren't any established standards or recommendations for employing thermal imaging to identify deception, it might be difficult to create consistent and trustworthy procedures. Finally, thermal imaging can be used as an additional tool in conjunction with other methods to produce more data for analysis [25] (Figure 4).
Figure 4. Thermal imaging [26]
2.5 Micro-expression method
It involves the systematic examination and interpretation of facial expressions to gain insights into an individual’s emotional state, intentions, and non-verbal communication. This analysis is based on the idea that human faces convey a wide range of emotions and subtle cues through changes in facial muscle movements and expressions. The advantage of this method is that it is considered to be an authentic indicator of someone’s mental state. It represents the person’s true feelings, as opposed to consciously controlled facial expressions, which may be used to conceal emotion. It is involuntary and typically occurs without the person’s awareness or control; it does not require physical contact or equipment attached to the person being tested. Micro-expressions are extremely brief, often lasting just a fraction of a second (as short as 1/25th of the second). This brevity can make them challenging to detect with the naked eye. The disadvantages are that human observers may introduce subjectivity into the analysis and different experts may reach different conclusions. Also, the interpretation of facial expression depends on the context, which can be complex to establish. In addition, facial expressions can vary across cultures interpreting challenges in cross-cultural situations (Table 2) [27].
According to technique analysis for multiple studies, the best method is micro-expressions for their advantages compared with the other techniques.
Table 2. Micro-expression method example [27]
|
AU |
Description |
Facial Muscle |
Example Image |
|
1 |
Inner brow raiser |
Frontalis, pars medialis |
|
|
2 |
Outer brow raiser |
Frontalis, pars lateralis |
|
|
4 |
Brow lowerer |
Corrugator supercilii, depressor supercillii |
|
|
5 |
Upper lid raiser |
Levalor palpebrae super ioris |
|
|
6 |
Cheek raiser |
Orbicularis oculi, pars orbitalis |
|
|
7 |
Lid tightener |
Orbicularis oculi, pars palpebralis |
|
|
9 |
Nose wrinkler |
Levator labil superioris alaqu ae nasi |
According to previous studies, micro-expression is the best method to detect lies [2], so new approaches and algorithms have been created, and the development of artificial intelligence has produced a blatant advancement in the realm of research [6]. There is a different way to detect lies using micro-expression. This paper provides the findings of studies that use images and videos, where all studies use the supervised method. Table 3 shows the studies that use images and videos. The table shows the algorithm that the research uses, the data set, the target, and the achievement.
Table 3. Micro-expression lie detection studies
|
Algorithm |
Field |
Dataset |
Target |
Achievement |
Ref. |
|
CNN and Random Forest classifier |
Video game |
Self-made dataset based on game |
Detect lies using AI |
Random forest outperformed CNN with this model's 86% accuracy. |
[28] |
|
OpenFace |
Evidence: Visual Lexical Acoustic |
Employ 61 fraudulent and 60 true video sources. |
Data-driven automated deceit detection |
Truth video has 81.1% accuracy, whereas deception video has 76.20%. It automatically recognized gestures and mapped textual analysis. CNN was less accurate than Random Forest at 86%. |
[29] |
|
Mathematical algorithm Video frames Embedded vision system |
Video |
Video made by a high-speed camera |
lie detection |
The expression recognition accuracy is 85%. |
[30] |
|
3D CNN |
Video |
Samples from: •SMIC •SAMM •CASM II |
extracted high-level features with details of micro-expressions. |
UAR is 0.7605 and UF1 is o.7353, which exceeds the most current state-of-the-art models. |
[31] |
|
Classifiers of threshold-base |
Thermal images |
Interview video |
Crime investigation lie detection |
Blind predictions yield 80% for a method that employs thermal imaging to assess the quantity of sweat on the face induced by stress. |
[32] |
|
Deep CNN and feed those characteristics to LSTM |
Video |
Video from a spontaneous database |
Evaluate the prosed network model TLCNN |
TLCNN is better than some state-of-art |
[33] |
|
HAAR Cascade algorithm in Matlab |
Video image |
Create continuous video using a Mathlab-connected camera. |
•Detect lie •Drivers test |
When the flash pattern matched the target contrast pattern, the person in question was determined to be lying. |
[34] |
|
Random Forest |
Video |
Open-source datasets like YouTube |
Forensic interview to identify four treatments (eye blink, eyebrow motion, wrinkle incidence, mouth motion). |
Use computer vision to detect high-risk fraud using facial vale |
[35] |
|
Utilizing approaches based on 2D appearance to describe 3D face features |
TV program (video) |
non-public dataset of real (genuine) and posed (deceptive) face expressions. |
Lie detection |
Computer vision methods provide 76.92% accuracy in facial signals for lying detection in high-stakes situations. |
[36] |
|
Underlying 2-layer GRU DepLie |
Video |
Kaggle has FER-2013 faces with emotions and humans talking truth or lies video |
Lie detection |
The validation set's correct classification rat (CCR) is 81.82%, the training set's 94.31% and 25 leave-one-out cross-validation trials get it to 100%. |
[37] |
|
AUS |
Video and computer vision |
satisfied and spontaneously dataset. |
•Security national •Clinical field |
Micro-expressions intensity Emotion marking standards Rapid facial expressions |
[38] |
|
MVL |
Video |
Public multimedia sources from TV interviews and court trials |
Deception detection trouble |
Multi-view learning deception detection increases classification above current methods. |
[39] |
|
Self-supervised learning with MobileNet2 |
Online lecture |
CelebA(available on the net for free) for pre-training Self-made for testing |
A multimodal system can detect liars |
an AI strategy to help instructors address this developing problem for a fairer learning environment |
[40] |
|
Tow classification algorithm: Decision trees Random FOREST |
Video |
Public court video dataset |
Compear between the two algorithm |
found Random Forest is better than Decision trees |
[41] |
|
Shallow CNN |
Image |
Use five open datasets: •SAMM •CASMEII •CASME •FERPlus •FER2013 |
Detect micro-expressions: Lie, Pain, and Babysitting |
aSHCNN architecture without temporal can detect static expressions without huge datasets and provide increased saliency map |
[42] |
|
HAAR-LIK feature-based cascade classifier |
Video |
Interview video |
Indicating by psychological |
The result shows excellent nonverbal behavior detection. |
[9] |
|
Facial emotion recognition using convolutional neural network |
Image |
Extended Cohn-Kanade Caltech faces CMU NIST |
Detect lie Learn based on mood |
The accuracy of FERC is about 96%. |
[43] |
|
Deep Tiefes FCNN |
Image |
Kaggle dataset |
Deception detection |
Tiefes FCNN for micro facial expression: 99.02% accuracy, 98.82% precision, 97.8% F1-score, 56.31 PSNR, 96.31 CC. |
[44] |
|
Multi-model detector |
Video |
Real-life video |
Lie detection |
It uses high-level and low-level visual characteristics to predict dishonesty independent of identification. |
[45] |
|
Face reader |
Video |
Videotape |
•Child lie detection •Psychology forensic implications |
automatically decipher the lying child's face and correlate it with lie detection. |
[46] |
|
Shi-Tomasi corner |
Image |
Collected data from 30 various scenarios and subject |
Interrogaitions rooms Airports |
The combined thermal and visual classifiers achieved an accuracy of 61.74% and 59.73%, outperforming physiological characteristics by 53.02% and reducing error rates by 18.56% and 14.28% respectively. |
[47] |
|
Micro-expression training tool (METT) |
Video |
Different types of video |
Improve detect lie |
The METT group did not outperform those in the bogus untraining and training groups when using Bayesian analyses. |
[48] |
|
Kanade-Lucas-Tomasi algorithm |
Video |
Use two datasets: •CASMEII •SMIC |
Propose a local binary pattern and optical flow histogram approach for using MEs on huge videos. |
automated micro-expression, system analysis identifies MEs. |
[49] |
|
Lightweight pre-trained CNN call ResNet 18 |
Image |
FER-2013 dataset |
Detect lie |
Detect the fear from the liar's face |
[50] |
The limitations the researchers found in the previous studies are that users are not professionals in lie detection, the amount of dataset is small, and not every dataset is public for the users. Finally, the way to record video and get pictures is clear; in other words, uncontrolled facts like head pose, illumination, and facial occlusion.
Although analyzing facial microexpressions is important for lie detection, it raises ethical concerns and privacy issues because it can be misused in different ways. Microexpressions, which are involuntary and fleeting, promise to reveal individuals' true feelings even when they try to hide them, thus violating emotional privacy [51, 52]. Artificial intelligence techniques such as deep learning in micro-expression recognition lead to the detection of real emotions, which violates personal rights as it makes emotional displays. This highlights the need to address the ethical implications and prevent potential misuse of this technology [53]. These concerns are particularly important in scenarios such as national security, clinical diagnosis, interrogations, and business negotiations where partial expression analysis is applied [8]. According to what was mentioned above, it is necessary to educate researchers about the necessity of preserving privacy and warn them of the resulting legal consequences and penalties, as well as educate individuals about the necessity of taking pledges from data set collectors not to use their data except for scientific purposes and to preserve their privacy. Finally, maintaining the confidentiality of data and using these technologies for humanitarian purposes and to serve society.
This paper shows the reason for using and developing lie detection and the techniques used to do that, and then it focuses on the best technique and shows what researchers do to develop the way to detect lies using facial expressions by showing the methods, dataset, field that the researcher uses either image or video, target, achievement, and finally the limitation that the researchers found in their research, which may be scientific, technical and ethical limitations. This paper exploited its presentation of these limitations to help researchers deal with, treat, and solve them by finding new methods or techniques or choosing a different dataset. Artificial intelligence plays an effective role in enhancing the lie detection process, and its multiple techniques and methods are considered the most important for effective detection in the micro expression method. However, the clarity and accuracy of the source, whether it is a video or an image, significantly affect the results, so this paper advises researchers in this field to take this point into account by enhancing image and/ or video quality using preprocessing techniques, in addition to providing a suitable data set. Finally improving the user of the lie detection method through effective and accurate training will significantly enhance the detection results.
[1] Vrij, A., Verschuere, B. (2013). Lie Detection in a Forensic Context. Oxford University Press. https://doi.org/10.1093/OBO/9780199828340-0122
[2] Saini, R., Rani, P. (2022). LDM: A systematic review on lie detection methodologies. Preprints.org. https://doi.org/10.20944/preprints202212.0443.v1
[3] Oswald, M. (2020). Technologies in the twilight zone: early lie detectors, machine learning and reformist legal realism. International Review of Law, Computers & Technology, 34(2): 214-231. https://doi.org/10.1080/13600869.2020.1733758
[4] Abd, S.H., Hashim, I.A., Jalal, A.S.A. (2021). Hardware implementation of deception detection system classifier. Periodicals of Engineering and Natural Sciences, 10(1): 151-163. http://doi.org/10.21533/pen.v10i1.2594
[5] Sousedikova, L., Hromada, M., Adamek, M. (2021). Analysis of artificial intelligence lie detector developed for airport security. Tomas Bata University in Zlin.
[6] Jupe, L.M., Keatley, D.A. (2020). Airport artificial intelligence can detect deception: or am i lying? Security Journal, 33(4): 622-635. https://doi.org/10.1057/s41284-019-00204-7
[7] Sousedikova, L., Malatinsky, A., Drofova, I., Adamek, M. (2021). The role of lie detection based system in controlling borders. In Annals of DAAAM and Proceedings of the International DAAAM Symposium, pp. 384-388. https://doi.org/10.2507/32nd.daaam.proceedings.056
[8] Oravec, J.A. (2022). The emergence of “truth machines”?: Artificial intelligence approaches to lie detection. Ethics and Information Technology, 24(1): 6. https://doi.org/10.1007/s10676-022-09621-6
[9] Setiawan, A.B., Anwar, K., Azizah, L., Prahara, A. (2019). Real-time facial expression recognition to track non-verbal behaviors as lie indicators during interview. Signal and Image Processing Letters, 1(1): 25-31. https://doi.org/10.31763/simple.v1i1.144
[10] Bergström, E. (2023). To spy the lie. Detecting the insider threat of espionage. Master thesis, Stockholm University.
[11] Walczyk, J.J., Schwartz, J.P., Clifton, R., Adams, B., Wei, M.I.N., Zha, P. (2005). Lying person-to-person about life events: A cognitive framework for lie detection. Personnel Psychology, 58(1): 141-170. https://doi.org/10.1111/j.1744-6570.2005.00484.x
[12] Khan, W., Crockett, K., O'Shea, J., Hussain, A., Khan, B. M. (2021). Deception in the eyes of deceiver: A computer vision and machine learning based automated deception detection. Expert Systems with Applications, 169: 114341. https://doi.org/10.1016/j.eswa.2020.114341
[13] Zanette, S., Gao, X., Brunet, M., Bartlett, M.S., Lee, K. (2016). Automated decoding of facial expressions reveals marked differences in children when telling antisocial versus prosocial lies. Journal of Experimental Child Psychology, 150: 165-179. https://doi.org/10.1016/j.jecp.2016.05.007
[14] D’Ulizia, A., D’Andrea, A., Grifoni, P., Ferri, F. (2023). Detecting deceptive behaviours through facial cues from videos: A systematic review. Applied Sciences, 13(16): 9188. https://doi.org/10.3390/app13169188
[15] Thannoon, H.H., Ali, W.H., Hashim, I.A. (2018). Detection of deception using facial expressions based on different classification algorithms. In 2018 Third Scientific Conference of Electrical Engineering (SCEE), Baghdad, Iraq, 2018, pp. 51-56. https://doi.org/10.1109/SCEE.2018.8684170
[16] Brennen, T., Magnussen, S. (2023). Lie detection: What works? Current Directions in Psychological Science, 32(5): 395-401. https://doi.org/10.1177/09637214231173095
[17] Vicianova, M. (2015). Historical techniques of lie detection. Europe's Journal of Psychology, 11(3): 522-534. https://doi.org/10.5964/ejop.v11i3.919
[18] Budaházi, Á.D. (2023). Polygraph test–with or without? In Thematic Conference Proceedings of International Significance, pp. 155-171. http://eskup.kpu.edu.rs/dar/article/view/406.
[19] Yu, R., Wu, S.J., Huang, A., Gold, N., Huang, H., Fu, G., Lee, K. (2019). Using polygraph to detect passengers carrying illegal items. Frontiers in Psychology, 10: 419536. https://doi.org/10.3389/fpsyg.2019.00322
[20] Awan, S.N., Roy, N., Jiang, J.J. (2010). Nonlinear dynamic analysis of disordered voice: The relationship between the correlation dimension (D2) and pre-/post-treatment change in perceived dysphonia severity. Journal of Voice, 24(3): 285-293. https://doi.org/10.1016/j.jvoice.2008.11.003
[21] Cósetl, R.C., López, J.D.B. (2011). Voice Stress Detection: A method for stress analysis detecting fluctuations on Lippold microtremor spectrum using FFT. In CONIELECOMP 2011, 21st International Conference on Electrical Communications and Computers, San Andres Cholula, Mexico, pp. 184-189. https://doi.org/10.1109/CONIELECOMP.2011.5749357
[22] Bles, M., Haynes, J.D. (2020). Detecting concealed information using brain-imaging technology. Neuroscience and Crime, pp. 82-92. https://doi.org/10.1080/13554790801992784
[23] Kholodny, Y.I., Kartashov, S.I., Malakhov, D.G., Orlov, V.A. (2021). Improvement of the technology of fMRI experiments in the concealed information paradigm. Brain-Inspired Cognitive Architectures for Artificial Intelligence: BICA*AI 2020, 591-597. https://doi.org/10.1007/978-3-030-65596-9_73
[24] MRI Lie Detectors - IEEE Spectrum. https://spectrum.ieee.org/mri-lie-detectors.
[25] Jain, U., Tan, B., Li, Q. (2012). Concealed knowledge identification using facial thermal imaging. In 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Kyoto, Japan, pp. 1677-1680. https://doi.org/10.1109/ICASSP.2012.6288219
[26] Gołaszewski, M., Zajac, P., Widacki, J. (2015). Thermal Vision as a method of detection of deception: A review of experiences. European Polygraph, 9(1): 5-24. https://doi.org/10.1515/ep-2015-0001
[27] M. Yan. (2019). What is facial expression?. https://medium.com/@michelleyan_75844/what-is-facial-expression-767ba76a9b2b.
[28] Haudenschild, J., Kafadar, V.A., Tschakert, H., Ullmann, L. (2022). Lie detector AI: Detecting lies through fear. Empathic AI / Affective Computing Project Report.
[29] Jaiswal, M., Tabibu, S., Bajpai, R. (2016). The truth and nothing but the truth: Multimodal analysis for deception detection. In 2016 IEEE 16th International Conference on Data Mining Workshops (ICDMW), Barcelona, Spain, pp. 938-943. https://doi.org/10.1109/ICDMW.2016.0137
[30] Owayjan, M., Kashour, A., Al Haddad, N., Fadel, M., Al Souki, G. (2012). The design and development of a lie detection system using facial micro-expressions. In 2012 2nd International Conference on Advances in Computational Tools for Engineering Applications (ACTEA), Beirut, Lebanon, pp. 33-38. https://doi.org/10.1109/ICTEA.2012.6462897
[31] Liong, S.T., Gan, Y.S., See, J., Khor, H.Q., Huang, Y.C. (2019). Shallow triple stream three-dimensional cnn (ststnet) for micro-expression recognition. In 2019 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2019), Lille, France, pp. 1-5. https://doi.org/10.1109/FG.2019.8756567
[32] Dcosta, M., Shastri, D., Vilalta, R., Burgoon, J.K., Pavlidis, I. (2015). Perinasal indicators of deceptive behavior. In 2015 11th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG), Ljubljana, Slovenia, pp. 1-8. https://doi.org/10.1109/FG.2015.7163080
[33] Wang, S.J., Li, B.J., Liu, Y.J., et al. (2018). Micro-expression recognition with small sample size by transferring long-term convolutional neural network. Neurocomputing, 312: 251-262. https://doi.org/10.1016/j.neucom.2018.05.107
[34] Singh, B., Rajiv, P., Chandra, M. (2015). Lie detection using image processing. In 2015 International Conference on Advanced Computing and Communication Systems, Coimbatore, India, pp. 1-5. https://doi.org/10.1109/ICACCS.2015.7324092
[35] Su, L., Levine, M.D. (2014). High-stakes deception detection based on facial expressions. In 2014 22nd International Conference on Pattern Recognition, Stockholm, Sweden, pp. 2519-2524. https://doi.org/10.1109/ICPR.2014.435
[36] Su, L., Levine, M. (2016). Does “lie to me” lie to you? An evaluation of facial clues to high-stakes deception. Computer Vision and Image Understanding, 147: 52-68. https://doi.org/10.1016/j.cviu.2016.01.009
[37] Feng, K.J. (2023). DeepLie: Detect lies with facial expression (computer vision). https://cs230.stanford.edu/projects_spring_2021/reports/0.pdf.
[38] Yan, W.J., Wu, Q., Liu, Y.J., Wang, S.J., Fu, X. (2013). CASME database: A dataset of spontaneous micro-expressions collected from neutralized faces. In 2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG), Shanghai, China, pp. 1-7. https://doi.org/10.1109/FG.2013.6553799
[39] Carissimi, N., Beyan, C., Murino, V. (2018). A multi-view learning approach to deception detection. In 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2018), Xi'an, China, pp. 599-606. https://doi.org/10.1109/FG.2018.00095
[40] Duong, D.C., Nguyen, N.N., Tam, N.D. (2022). A deep learning powered system to lie detection while online study. Traitement du Signal, 39(3): 893-898. https://doi.org/10.18280/ts.390314
[41] Pérez-Rosas, V., Abouelenien, M., Mihalcea, R., Burzo, M. (2015). Deception detection using real-life trial data. In Proceedings of the 2015 ACM on International Conference on Multimodal Interaction, Seattle Washington, USA, pp. 59-66. https://doi.org/10.1145/2818346.2820758
[42] Miao, S., Xu, H., Han, Z., Zhu, Y. (2019). Recognizing facial expressions using a shallow convolutional neural network. IEEE Access, 7: 78000-78011. https://doi.org/10.1109/ACCESS.2019.2921220
[43] Mehendale, N. (2020). Facial emotion recognition using convolutional neural networks (FERC). SN Applied Sciences, 2(3): 446. https://doi.org/10.1007/s42452-020-2234-1
[44] Durga, B.K., Rajesh, V., Jagannadham, S., Kumar, P.S., Rashed, A.N.Z., Saikumar, K. (2023). Deep learning-based micro facial expression recognition using an adaptive tiefes fcnn model. Traitement du Signal, 40(3): 1035-1043. https://doi.org/10.18280/ts.400319
[45] Wu, Z., Singh, B., Davis, L., Subrahmanian, V. (2018). Deception detection in videos. Proceedings of the AAAI Conference on Artificial Intelligence, 32(1): 1695-1702. https://doi.org/10.1609/aaai.v32i1.11502
[46] Gadea, M., Aliño, M., Espert, R., Salvador, A. (2015). Deceit and facial expression in children: the enabling role of the “poker face” child and the dependent personality of the detector. Frontiers in Psychology, 6: 1089. https://doi.org/10.3389/fpsyg.2015.01089
[47] Abouelenien, M., Mihalcea, R., Burzo, M. (2016). Analyzing thermal and visual clues of deception for a non-contact deception detection approach. In Proceedings of the 9th ACM International Conference on PErvasive Technologies Related to Assistive Environments, Corfu Island, Greece, pp. 5. https://doi.org/10.1145/2910674.2910682
[48] Jordan, S., Brimbal, L., Wallace, D.B., Kassin, S.M., Hartwig, M., Street, C. N. (2019). A test of the micro-expressions training tool: Does it improve lie detection?. Journal of Investigative Psychology and Offender Profiling, 16(3): 222-235. https://doi.org/10.1002/jip.1532
[49] Li, X., Hong, X., Moilanen, A., Huang, X., Pfister, T., Zhao, G., Pietikäinen, M. (2017). Towards reading hidden emotions: A comparative study of spontaneous micro-expression spotting and recognition methods. IEEE Transactions on Affective Computing, 9(4): 563-577. https://doi.org/10.1109/TAFFC.2017.2667642
[50] Al Salti, S.M.S., Sherimon, V., Remya, R.K. (2022). An artificial intelligence lie detection model using ResNet-18 Network. International Journal of Engineering Research & Technology, 11(9): 199-202.
[51] Zhao, G., Li, X., Li, Y., Pietikäinen, M. (2023). Facial Micro-expressions: An overview. Proceedings of the IEEE, 111(10): 1215-1235. https://doi.org/10.1109/JPROC.2023.3275192
[52] Zhang, H., Zhang, H. (2022). A review of micro-expression recognition based on deep learning. In 2022 International Joint Conference on Neural Networks (IJCNN), Padua, Italy, pp. 1-8. https://doi.org/10.1109/IJCNN55064.2022.9892307
[53] Low, Y.Y., Tanvy, A., Phan, R.C.W., Chang, X. (2022). AdverFacial: Privacy-preserving universal adversarial perturbation against facial micro-expression leakages. In ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing, Singapore, pp. 2754-2758. https://doi.org/10.1109/ICASSP43922.2022.9746848
