Pith. sign in

REVIEW 6 major objections 6 minor 66 references

Real-Time Confidence Detection through Facial Expressions and Hand Gestures

T0 review · 6 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that a webcam-only system can read a speaker's confidence in real time from six nonverbal cues combined by fixed weights, reaching 90% accuracy in tests.

desk verdict The prototype may work as a demo, but the 90% accuracy claim is never measured against any ground truth, and the paper is undercut by internal inconsistencies and fabricated references. read the letter →

arxiv 2506.09153 v1 pith:6KVILUTL submitted 2025-06-10 cs.HC

Tanjil Hasan Sakib , Samia Jahan Mojumder , Rajan Das Gupta , Md Imrul Hasan Showmick , Md. Yeasin Rahat , Md. Jakir Hossen This is my paper
classification cs.HC
keywords real-timeconfidencedetectionfacialexpressionshandgesturesheadposeestimationMediaPipeFaceMeshEuleranglesweightedscoringmodelengagementmonitoring
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper sets out to show that a webcam alone can estimate a speaker's confidence in real time, by folding six visible cues into one running score: smiles, lip movement, blink rate, head orientation, gaze steadiness, and hand gestures. The prototype uses MediaPipe's Face Mesh and Hands models to extract landmarks, computes Euler angles (yaw, pitch, roll) from the face mesh to track head direction, and applies a weighted combination of the six cues. On two-minute webcam speeches from ten participants, the authors report an average accuracy of 90% and head-orientation recognition at the same accuracy up to four feet away. The payoff, if the result holds, is a low-cost way to monitor engagement and confidence in online interviews, virtual classrooms, and remote meetings without special hardware.

What carries the argument

The load-bearing mechanism is a weighted scoring model built on MediaPipe landmark outputs. MediaPipe Face Mesh supplies 3D facial landmarks from which Euler angles — yaw, pitch, and roll — are computed for head orientation, while MediaPipe Hands tracks hand motion. Each of the six features is mapped to a sub-score through hand-set thresholds: a smile is detected when the lip-aspect ratio exceeds 1.5, blink rate above 15 per minute lowers the score, head deviation beyond ±10 degrees lowers it, and hand speeds of 0.2–0.5 m/s count as confident. The weighted sum of these sub-scores, computed at roughly 30 ms per frame, is the confidence percentage.

What would settle it

Collect webcam recordings from a larger, more diverse group of speakers, run the system on every frame, and compare its scores with independent human ratings of confidence; if agreement is close to chance, or if a speaker can flip a high score to low simply by smiling while gesturing quickly, the fixed-weight claim collapses.

Watch

Extended reading notes

Core claim

The paper's central claim is that confidence during virtual interaction can be read automatically from nonverbal behaviour alone. Each frame is reduced to six features, each feature is thresholded to produce a confidence sub-score, and the sub-scores are combined with fixed weights — hand gestures 30%, head movement 15%, smile, lip movement, blink rate, and gaze 10% each — to produce a percentage shown to the user. The authors report that speakers with moderate hand speed, frequent smiles, steady gaze, active lip movement, and normal blink rates scored in the high range (0.9–1.2), while excessive blinking, erratic head movements, and rapid gestures scored below 0.5. This weighted score is the system's real-time confidence estimate.

Load-bearing premise

The load-bearing premise is that a fixed linear combination of six visible cues, with thresholds that are the same for everyone, captures what confidence looks like across people and settings.

Editorial extensions

If this is right

  • A system that runs on an ordinary webcam could add live confidence feedback to existing online meeting, interview, and classroom software without extra sensors.
  • Head-orientation tracking that holds 90% accuracy at four feet would let engagement be monitored when a speaker is not close to the camera.
  • Combining face, head, and hand signals gives the score redundant evidence, so a cue that is ambiguous on its own can still contribute through the others.
  • The same feature pipeline can be pointed at adjacent tasks such as proctoring, where gaze and head shifts are already treated as attention signals.
  • The authors' stated next steps — voice analysis and multi-user tracking — would extend the artifact from a single-speaker tool to richer interaction settings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The fixed thresholds and weights are likely person- and culture-dependent, so a version that learns the mapping from human-labeled data would probably generalize better than the hand-set linear combination.
  • The text says effectiveness will be judged against human ratings but does not report that comparison, leaving external validity as the natural next check.
  • The same landmarks could support other social-signal estimates, such as nervousness, dominance, or rapport, because they are not specific to confidence.
  • If the score tracks perceived confidence, the system could be turned into a live coaching aid that tells speakers when their delivery reads as hesitant.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

6 major / 6 minor

Summary. The manuscript describes a real-time system that uses MediaPipe Face Mesh and Hands to track facial landmarks, head pose (Euler angles), gaze, blink rate, lip movement, and hand gestures, and combines them into a weighted confidence score intended for virtual interactions such as online interviews and remote meetings. The authors claim a 90% accuracy for the system, including for head-orientation identification at up to four feet, and report category-level percentages about participants' behavior. The central claim is presented in the abstract and in Sections IV.A and V.F. The paper does not, however, define or compute any accuracy metric against a ground truth, and the promised human-judgment comparison in Section IV.B is not reported.

Significance. If the 90% accuracy claim were properly supported, the contribution would be relevant to affective computing and web-based HCI, since the system integrates multiple inexpensive webcam-based cues in real time. I credit the authors for making the design-science process explicit and for specifying a modular MediaPipe-based pipeline with 30 ms/frame processing and a clearly enumerated feature set. The underlying technologies (face mesh landmarks, Euler-angle head-pose estimation, blink and lip-movement ratios) are plausible building blocks. However, the evaluation is missing: there is no external ground truth, no defined accuracy metric, no statistical analysis, and no comparison with human ratings. The reported numerical claims are therefore not evidence for the stated performance, and the paper's significance cannot be assessed beyond the prototype implementation.

major comments (6)
  1. [IV.B / V.F] Section IV.B states that effectiveness is evaluated by comparing system outputs with human judgments of confidence in controlled experiments, but Section V.F, titled "Correlation with Human Evaluation," reports only a single participant's live score (90.77%) from Figure 6. No correlation coefficient, confusion matrix, inter-rater statistic, or other comparison with human ratings is reported. The promised ground-truth evaluation is absent, so the 90% accuracy claim has no measured basis.
  2. [Abstract / IV.A / V.F] The "90% accuracy" figure is never operationalized. It could refer to head-orientation classification (as in the abstract), agreement with human confidence ratings, or the live score percentage in Figure 6, but no metric is defined and no ground-truth labels are described. The live score shown in Figure 6 is a system output, not an accuracy measure. Any central claim of 90% accuracy requires a defined comparison target and a test protocol; neither is provided.
  3. [V.C / Table 1] The six weights in Table 1 sum to 85% (30+10+10+10+15+10), not 100%, and no normalization is described in Section V.C or V.E. As a result, the weighted confidence score is not well defined as stated; if the weights are meant to be fractions of a total, the missing normalization changes every reported score.
  4. [V.A] The participant distributions in Section V.A are arithmetically inconsistent: hand gestures (70%+25%+25%=120%), smile (60%+25%+30%=115%), and head movement (55%+25%+40%=120%) exceed 100%. These sums are impossible for mutually exclusive categories and indicate that the quantitative descriptions do not correspond to a coherent data set.
  5. [V.A / V.C] The thresholds (smile ratio >1.5, blink rate >15/min, head deviation ±10°, hand speed 0.2–0.5 m/s) and the factor weights are introduced without justification or sensitivity analysis. Since the confidence score is constructed from these thresholds and weights, the later finding that these features "indicate" confidence is circular unless validated against an independent ground truth such as human confidence ratings. No such validation is reported.
  6. [V.A / IV.C / V.D] The study reports only ten participants and no error bars, confidence intervals, or statistical tests. Percentages such as "70% of participants scored above 0.9" are not accompanied by sample sizes per category or any measure of variance, and the "average accuracy of 90%" in Section IV.A therefore cannot be assessed.
minor comments (6)
  1. [V / VI] Section V and Section VI are both titled "Results and Analysis"; Section VI is actually a discussion and future-work section. The duplicate numbering and headings should be fixed.
  2. [V.E] The "Weighting Calculation" heading in Section V.E appears at the start of what is actually a findings list, not a weighting calculation; the subsection structure should be reorganized.
  3. [IV.C] "Conventional Neural Networks" should presumably be "Convolutional Neural Networks."
  4. [V.A] There is a typo, "ndicative," and multiple hyphenation/line-break artifacts (e.g., "be -comes," "-time") in the supplied text; the manuscript needs copyediting.
  5. [References] Several references appear to be placeholder entries (e.g., [12], [13], [14] with generic author names and synthetic DOIs), and one entry [12] has a duplicated "[12]" in the bibliography; these should be replaced with verifiable sources.
  6. [Figure 6] The text around Figure 6 gives conflicting score values: the figure caption reports 90.77% while the body text says "a 90.00% confidence score"; these should be reconciled.

Circularity Check

2 steps flagged · score 7.0 of 10

The central 90% accuracy claim is the system's own confidence score, not a measured accuracy against any ground truth.

  1. fitted input called prediction [Abstract; Section IV.A; Section V.F]
    "The system was tested with an average accuracy of 90% ... Figure 6 shows a real-time confidence detection system using Mediapipe to track facial and hand landmarks, evaluating gestures, smiles, blink rates, and head movements, and providing a confidence score (e.g., 90.77%)"

    The 'average accuracy of 90%' is supported only by displaying the system's own computed confidence score (90.77%). No comparison with human confidence judgments, known head orientations, or any other ground-truth label is reported, although Section IV.B promised evaluation by comparison with human judgments in controlled experiments. The confidence score is produced by the hand-weighted scoring rule in Section V.C and Table 1, so the reported 'accuracy' is the output of the scoring function, not a measured agreement with an external criterion. The central claimed prediction thus reduces to the system's own output by construction.

  2. self definitional [Section V.A and Section V.C / Table 1]
    "Moderate movements between 0.2–0.5 m/s correlated with high confidence scores (0.9–1.2) ... Smiling increased confidence scores by a factor of 1.2 ... if the gaze confidence score is 0.9 and its assigned weight is 15%, its contribution to the total score would be 0.9 × 0.15 = 0.135."

    The reported 'correlations' between gestures and confidence are not empirical findings: the scoring rule itself assigns high scores to these feature ranges through hand-selected thresholds, smile multipliers, and Table 1 weights. Saying that participants with moderate gestures 'scored above 0.9' is a restatement of the scoring function, not an independent test of the gesture-confidence relationship. Consequently, the confidence score—and the 90% accuracy derived from it—is defined by the system's own ad hoc rules rather than validated against external labels.

full rationale

The paper's main quantitative claim, 90% accuracy, is not anchored to any external ground truth. Section IV.A asserts 'The system was tested with an average accuracy of 90%', and the abstract repeats a 90% head-orientation accuracy, but the only supporting exhibit is Figure 6, which displays a live confidence score of 90.77% computed by the system itself. Section IV.B promises that effectiveness 'is evaluated by comparing the system's outputs with human judgments of confidence in controlled experiments', yet no such comparison, correlation coefficient, confusion matrix, or inter-rater agreement is ever reported. Section V.F, headed 'Correlation with Human Evaluation', contains only a description of the system's own output. Thus the central accuracy number reduces to the weighted scoring function in Section V.C and Table 1: the system reports a confidence percentage, and that same percentage is presented as accuracy. The secondary circularity is in Section V.A, where feature ranges are first assigned confidence values by ad hoc thresholds and multipliers, and then those assignments are reported as observed correlations with confidence. This is not a case of self-citation being load-bearing; references [56]–[60] are peripheral and do not carry the derivation. The MediaPipe-based landmark tracking and Euler-angle computation are legitimate components with independent support, but the paper's own promised evaluation is absent, so the 'prediction' of 90% accuracy is by construction the system's own score rather than a measured result. This warrants a score of 7: the central claim is substantially self-referential, though the underlying sensing pipeline has some independent content.

Assumptions & free parameters 11 free parameters · 3 assumptions · 0 invented entities

The central claim depends on a set of hand-picked weights and thresholds that are not derived from data. The system's 'accuracy' is not grounded in external labels or a validation set. No new entities are introduced.

free parameters (11)
  • weight_hand_gestures = 0.30
    Table 1 assigns 30% to hand gestures without documented optimization or data fit.
  • weight_smile = 0.10
    Table 1 assigns 10% to facial expressions (smile) without documented optimization.
  • weight_lip_movement = 0.10
    Table 1 assigns 10% to lip movement without documented optimization.
  • weight_blink_rate = 0.10
    Table 1 assigns 10% to blink rate without documented optimization.
  • weight_head_movement = 0.15
    Table 1 assigns 15% to head movement without documented optimization.
  • weight_gaze = 0.10
    Table 1 assigns 10% to gaze confidence without documented optimization.
  • smile_ratio_threshold = 1.5
    Section V.A uses a lip aspect ratio threshold of 1.5 to detect smiles; the threshold appears chosen by hand.
  • blink_rate_threshold = 15 blinks per minute
    Section V.A sets a blink rate exceeding 15 blinks per minute as an indicator of stress; the cutoff is stated without derivation.
  • head_deviation_threshold = 10 degrees
    Section V.A treats head deviations beyond plus or minus 10 degrees as reducing confidence; the threshold is ad hoc.
  • hand_speed_moderate_range = 0.2 to 0.5 m/s
    Section V.A defines moderate hand gesture speed as 0.2 to 0.5 m/s and associates it with high confidence; the range is chosen by hand.
  • smile_boost_factor = 1.2
    Section V.A states smiling increases confidence scores by a factor of 1.2; the multiplier is not fitted to data.
assumptions (3)
  • domain assumption Confidence can be quantified as a scalar weighted sum of independent non-verbal cues.
    The entire scoring architecture assumes linear additivity and independence of hand, smile, lip, blink, head, and gaze signals, with no justification or empirical support.
  • domain assumption MediaPipe landmark detections are sufficiently accurate for this task.
    The system's inputs depend entirely on MediaPipe face mesh and hand landmarks; no validation of landmark errors is provided in the paper.
  • domain assumption Ten webcam participants are representative of the target population.
    Data comes from ten participants giving two-minute speeches in an unspecified setting, with no diversity or controlled conditions, yet the accuracy claim is generalized.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Real-Time Confidence Detection through Facial Expressions and Hand Gestures." pith.science (2026). https://pith.science/paper/6KVILUTL

@misc{pith2026250609153,
  author       = {Pith},
  title        = {Pith review of: Real-Time Confidence Detection through Facial Expressions and Hand Gestures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6KVILUTL}},
  note         = {Machine review of arXiv:2506.09153}
}
read the original abstract

Real-time face orientation recognition is a cutting-edge technology meant to track and analyze facial movements in virtual environments such as online interviews, remote meetings, and virtual classrooms. As the demand for virtual interactions grows, it becomes increasingly important to measure participant engagement, attention, and overall interaction. This research presents a novel solution that leverages the Media Pipe Face Mesh framework to identify facial landmarks and extract geometric data for calculating Euler angles, which determine head orientation in real time. The system tracks 3D facial landmarks and uses this data to compute head movements with a focus on accuracy and responsiveness. By studying Euler angles, the system can identify a user's head orientation with an accuracy of 90\%, even at a distance of up to four feet. This capability offers significant enhancements for monitoring user interaction, allowing for more immersive and interactive virtual ex-periences. The proposed method shows its reliability in evaluating participant attentiveness during online assessments and meetings. Its application goes beyond engagement analysis, potentially providing a means for improving the quality of virtual communication, fostering better understanding between participants, and ensuring a higher level of interaction in digital spaces. This study offers a basis for future developments in enhancing virtual user experiences by integrating real-time facial tracking technologies, paving the way for more adaptive and interactive web-based platform.

Figures

Figures reproduced from arXiv: 2506.09153 by the authors.

Figure 1
Figure 1. presents the architecture of the real-time confidence detec- tion system. The process begins with capturing a real-time video stream, followed by frame capture and landmark detection to iden- tify key facial and hand features. Next, the facial landmark extraction stage analyzes face move- ment, gaze, blink rate, lip movement, mouth openness, and hand motion. These features are processed at 30ms per frame to compute … view at source ↗
Figure 2
Figure 2. A flowchart representing a proposed model for real [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Flowchart of the data collection process for real￾time confidence scoring. Data was collected from ten participants who delivered two￾minute speeches while being recorded via webcam, maintaining natural facial expressions and hand gestures. The recorded videos served as the foundation for analyzing key behavioral indicators of confidence, including smile detection, blink rate, head movement, hand gestures, lip movem… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: A flowchart illustrating the process of weighting and [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Flowchart showing the weighting and aggregation of [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 6
Figure 6. Figure 6: Participant uses his hand while talking, showing confidence at 90.77%. for virtual interviews or e-learning [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

66 extracted references · 59 canonical work pages

  1. [12]

    [12]Alex Doe and Lisa Huang. 2022. Multi-Face Recognition in Group Settings. IEEE Transactions on Image Processing 31, 4 (2022), 1234 –1245. doi:10.1109/TIP.2022.1234567

    arXiv 2022

  2. [13]

    John Doe and Jane Smith. 2024. Advancements in Emotion Classification via Facial and Body Gestures. Expert Systems 41, 5 (2024), e13759. doi:10.1111/exsy.13759

    work page doi:10.1111/exsy.13759 2024

  3. [1]

    Fahmid Al Farid, Noramiza Hashim, Junaidi Abdullah, Md Roman Bhuiyan, Wan Noor Shahida Mohd Isa, Jia Uddin, Mohammad Ahsanul Haque, and Mohd Nizam Husen. 2022. A structured and methodological review on vision - based hand gesture recognition system. Journal of Imaging 8, 6 (2022), 153

    work page 2022

  4. [2]

    Chris Anderson and Laura Green. 2024. Lip Movement Analysis for Speech Recognition. Speech Communication 150 (2024), 50 –65. doi:10.1016/j.specom. 2024.101234

    arXiv 2024

  5. [3]

    MCP Archana, CK Nitish, and Sandhya Harikumar. 2022. Real time face detection and optimal face mapping for online classes. In Journal of Physics: Conference Series, Vol. 2161. IOP Publishing, 012063

    work page 2022

  6. [4]

    Kiavash Bahreini, Rob Nadolski, and Wim Westera. 2016. Towards multimodal emotion recognition in e -learning environments. Interactive Learning Environ- ments 24, 3 (2016), 590–605

    work page 2016

  7. [5]

    Emily Brown and Raymond Chang. 2023. Gesture Recognition for Public Speaking and Leadership Training. Journal of Nonverbal Behavior 47, 6 (2023), 512–528. doi:10.1007/s10919-023-00322-9

    work page doi:10.1007/s10919-023-00322-9 2023

  8. [6]

    Badler, Mark Steedman, Brian Achorn, Bret Becket, Beverly Douville, Scott Prevost, and Matthew Stone

    Justine Cassell, Catherine Pelachaud, Norman I. Badler, Mark Steedman, Brian Achorn, Bret Becket, Beverly Douville, Scott Prevost, and Matthew Stone

    work page

Show all 66 references
  1. [7]

    Ginevra Castellano, Santi Dominguez Villalba, and Antonio Camurri. 2009. Au- tomatic Recognition of Emotions in Natural Hand Gestures. (2009), 126 – 135

    2009

  2. [8]

    Ayesha Chowdhury and Faisal Rahman. 2020. Face and eye movement monitoring for online interview integrity. In Proceedings of the 2020 International Conference on Intelligent Systems. IEEE, 75 –82

    2020

  3. [9]

    Arnaud Dapogny, Kevin Bailly, and Séverine Dubuisson. 2018. Confidence - weighted local expression predictions for occlusion handling in expression recog- nition and action unit detection. International Journal of Computer Vision 126 (2018), 255–271

    2018

  4. [10]

    Laslo Dinges, Marc-André Fiedler, Ayoub Al-Hamadi, Thorsten Hempel, Ahmed Abdelrahman, Joachim Weimann, and Dmitri Bershadskyy

  5. [11]

    Alice Doe and Bob Brown. 2023. Body Language Analysis in Healthcare: An Overview. Frontiers in Psychology 14 (2023), 9325107. doi:10.3389/fpsyg.2023.9325107

    2023

  6. [14]

    John Doe and Jane Smith. 2024. Recognition and Classification of Smiles using Computer Vision. IEEE Transactions on Affective Computing 15 (2024), 102 –

    2024

  7. [15]

    David Dukić and Ana Sovic Krzic. 2022. Real -time facial expression recogni - tion using deep learning with application in the active classroom environment. Electronics 11, 8 (2022), 1240

    2022

  8. [16]

    CG Espinosa Sandoval. 2019. Multiple face detection and recognition system design applying deep learning in web browsers. https://scholarworks.uark.edu/ csceuht/74/ University of Arkansas ScholarWorks

    2019

  9. [17]

    Mark Evans and Rachel Adams. 2024. Gaze Tracking: Techniques and Appli - cations. ACM Transactions on Multimedia Computing, Communications, and Applications 20, 1 (2024), 1–18. doi:10.1145/3578921

    2024 doi

  10. [18]

    Gao and Y

    L. Gao and Y. Xu. 2012. Face orientation recognition based on multiple facial feature triangles. In 2012 International Conference on Control Engineering and Communication Technology. 175. doi:10.1109/iccect.2012.175

    2012 doi

  11. [19]

    Jens-Uwe Garbas, Tobias Ruf, Matthias Unfried, and Anja Dieckmann. 2013. Towards robust real -time valence recognition from facial expressions for mar - Figure 5 illustrates the process of calculating the confidence factor by assigning weights to individual facial gestures, su...

    2013

  12. [20]

    Luis Garcia and Juan Solis. 2021. Cheating detection in online exams using face and eye tracking. In Proceedings of the 10th International Conference on Computer Science Education. IEEE, 215–220

    2021

  13. [21]

    Antonia F. de C. Hamilton. 2019. The Role of Eye Gaze During Natural Social Interactions in Typical and Autistic People. Frontiers in Psychology 10 (2019),

    2019

  14. [22]

    Alan R Hevner, Salvatore T March, Jinsoo Park, and Sudha Ram. 2004. Design science in information systems research. MIS quarterly (2004), 75 –105

    2004

  15. [23]

    Souta Hidaka, Tetsuo Nozawa, and Akihiro Yagi. 2010. Head and Eye Gaze Dynamics During Visual Attention Shifts in Complex Environments. Journal of Vision 10, 7 (2010), 562–562. doi:10.1167/10.7.562

    2010 doi

  16. [24]

    M Sazzad Hussain, Rafael A Calvo, and Fang Chen. 2014. Automatic cognitive load detection from face, physiology, task performance and fusion during affective interference. Interacting with computers 26, 3 (2014), 256 –268

    2014

  17. [25]

    Muhammad Nazrul Islam. 2017. Using a Design Science Research Approach in Human-Computer Interaction (HCI) Project: Experiences, Lessons and Future Directions. International Journal of Virtual and Augmented Reality (IJVAR) 1, 2 (2017), 42–59

    2017

  18. [26]

    Richard J. K. Jacob and Keith S. Karn. 2003. Eye Movements in Psychology: Methods, Data, and Theory. In The Mind’s Eye: Cognitive and Applied Aspects of Eye Movement Research. Elsevier, 573–605

    2003

  19. [27]

    Emily Johnson and Robert Brown. 2024. Eye Blink Rate Increases as a Function of Cognitive Load During an Auditory Task. Journal of Cognitive Neuroscience 36, 4 (2024), 675–690. doi:10.1162/jocn_a_01978

    2024 doi

  20. [28]

    Venkatesha K, Suma V, and Ravikumar G H. 2018. Real Time Emotion Recognition from Facial Images using Support Vector Machine. International Journal of Emerging Technologies and Innovative Research (JETIR) 5 (2018), 541–545. Issue 12. http://www.jetir.org/papers/JETIR1812593.pdf

    2018

  21. [29]

    Ashish Kapoor, Winslow Burleson, and Rosalind W Picard. 2007. Automatic prediction of frustration. International journal of human -computer studies 65, 8 (2007), 724–736

    2007

  22. [30]

    Sheheryar Khan, Guoxia Xu, Raymond Chan, and Hong Yan. 2017. An online spatio-temporal tensor learning model for visual tracking and its applications to facial expression recognition. Expert Systems with Applications 90 (2017), 427–438

    2017

  23. [31]

    Kim and Hana Nguyen. 2023. Interactive Visualization for Real -Time Feed - back in Confidence Detection Systems. Journal of Human -Computer Interaction 39, 4 (2023), 345–358. doi:10.1145/12345678

    2023 doi

  24. [32]

    Pavel Král and Ladislav Lenc. 2015. Confidence measure for experimental auto- matic face recognition system. In Agents and Artificial Intelligence: 6th Interna- tional Conference, ICAART 2014, Angers, France, March 6 -8, 2014, Revised Selected Papers 6. Springer, 362–378

    2015

  25. [33]

    Pranav Kumar, S L Happy, and Aurobinda Routray. 2016. A real -time robust facial expression recognition system using HOG features. In 2016 International Conference on Computing, Analytics and Security Trends (CAST). 289 –293. doi:10. 1109/CAST.2016.7914982

    2016

  26. [34]

    Ming Li and Jian Zhang. 2024. Analyzing Facial Features for Emotional Recognition. IEEE Transactions on Affective Computing 12, 2 (2024), 215 –

    2024

  27. [35]

    Juan Martinez and Priya Singh. 2022. VR and AR Applications in Confidence Building. Virtual Reality 26, 3 (2022), 243 –257. doi:10.1007/s10055 -021- 00505-1

    2022 doi

  28. [36]

    David Miller and Anna White. 2024. Hand Gesture Recognition: A Literature Review. International Journal of Human-Computer Interaction 40, 2 (2024), 200–220. doi:10.1080/10447318.2024.1986543

    2024

  29. [37]

    Sharmin Akter Milu, Azmath Fathima, Tanmay Talukder, Inzamamul Islam, and Md Ismail Siddiqi Emon. 2024. Design and Implementation of hand gesture detection system using HM model for sign language recognition development. Journal of Data Analysis and Information Processing 12, ...

    2024

  30. [38]

    Moutan Mukhopadhyay, Saurabh Pal, Anand Nayyar, Pijush Kanti Dutta Pra - manik, Niloy Dasgupta, and Prasenjit Choudhury. 2020. Facial emotion detection to assess Learner’s State of mind in an online learning system. In Proceedings of the2020 5th international conference on int...

    2020

  31. [39]

    Erik Murphy-Chutorian and Mohan M. Trivedi. 2009. Head Pose Estimation in Computer Vision: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence 31, 4 (2009), 607–626

    2009

  32. [40]

    Mehul Naik, Rohan Maloor, Shivam Pandey, and Dhiraj Amin. 2022. CONFI- DENCE LEVEL ESTIMATOR BASED ON FACIAL AND VOICE EXPRESSION RECOGNITION AND CLASSIFICATION. IRJET 9, 04 (2022)

    2022

  33. [41]

    Thien Nguyen and Wei Wang. 2020. Real -time eye gaze tracking for online learning applications. International Journal of Educational Technology 8, 3 (2020), 45–52

    2020

  34. [42]

    Madi Nuralin, Yevgeniya Daineko, Shadi Aljawarneh, Dana Tsoy, and Madina Ipalakova. 2024. The real -time hand and object recognition for virtual interaction. PeerJ Computer Science 10 (2024), e2110

    2024

  35. [43]

    Maja Pantic and Leon J. M. Rothkrantz. 2000. Automatic analysis of facial expressions: The state of the art. IEEE Transactions on pattern analysis and machine intelligence 22, 12 (2000), 1424–1445

    2000

  36. [44]

    Maja Pantic and Leon J. M. Rothkrantz. 2000. A Computational Model of Facial Expression Analysis for Detecting Emotional States. IEEE Transactions on Pattern Analysis and Machine Intelligence 22, 12 (2000), 1424 –1445

    2000

  37. [45]

    Wesley L Passos, Igor M Quintanilha, and Gabriel M Araujo. 2018. Real -time deep-learning-based system for facial recognition. Simpósio Brasileiro de Teleco- municações e Processamento de Sinais (SBrT) 37 (2018), 895 –899

    2018

  38. [46]

    Ken Peffers, Tuure Tuunanen, Marcus A Rothenberger, and Samir Chatterjee

  39. [47]

    Stefan Petridis, Pingchuan Ma, and Maja Pantic. 2017. Lip -Reading Based on Deep Neural Networks: A Review. (2017), 2867–2871

    2017

  40. [48]

    de Lange, Victor A

    Yair Pinto, Sander van Gaal, Floris P. de Lange, Victor A. F. Lamme, and Anil K. Seth. 2013. Using Eye Movements to Measure the Confidence of Seeing Conscious Perception. Consciousness and Cognition 22, 3 (2013), 729 –741

    2013

  41. [49]

    Keith Rayner. 1998. The Psychology of Eye Movements. Cognitive Psychology 15, 2 (1998), 145–180

    1998

  42. [50]

    Google Research. 2021. Mediapipe. Available at: https://google.github.io/ mediapipe/

    2021

  43. [51]

    Maria Rodriguez and Claire Johnson. 2021. Anti -cheating mechanisms in online learning: A review of current techniques. Journal of E -Learning and Higher Education 2021 (2021), 1–10

    2021

  44. [52]

    Anas Samara, Leo Galway, Raymond Bond, and Hui Wang. 2019. Affective state detection via facial expression analysis within a human –computer interaction context. Journal of Ambient Intelligence and Humanized Computing 10 (2019), 2175–2184

    2019

  45. [53]

    John Smith and Karen Lee. 2019. Improving face orientation detection for moni- toring online test -takers. IEEE Transactions on Learning Technologies 12, 4 (2019), 480–487

    2019

  46. [54]

    John Smith and Sarah Taylor. 2023. Voice Analysis for Confidence and Emotional States. Journal of Speech Processing 25, 7 (2023), 215 –230. doi:10.1234/jsp.2023.5678

    2023

  47. [55]

    Saeed Turabzadeh, Hongying Meng, Rafiq M Swash, Matus Pleva, and Jozef Juhar. 2018. Facial expression emotion detection for real -time embedded systems. Technologies 6, 1 (2018), 17

    2018

  48. [56]

    T., Morol, M

    Hosain, M. T., Morol, M. K., & Hossen, M. J. (2025). A hybrid self attentive linearized phrase structuredtransformer based RNN for financial sentenceanalysis with sentence level explainability

    2025

  49. [57]

    T., Abir, M

    Hosain, M. T., Abir, M. R., Rahat, M. Y., Mridha, M. F., & Mukta, S. H. (2024). Privacy preserving machine learning with federated personalized learning in artificially generated environment. IEEE open journal of the computer society

    2024

  50. [58]

    T., Zaman, A., Sajid, M

    Hosain, M. T., Zaman, A., Sajid, M. S., Khan, S. S., & Akter, S. (2023, October). Privacy preserving machine learning model personalization through federated personalized learning. In 2023 4th International Conference on Data Analytics for Business and Industry (ICDABI) (pp. 5...

    2023

  51. [59]

    Raihan, M., Saha, P.K., Gupta, R.D., Kabir, M.T., Tamanna, A.A., Harun -Ur- Rashid, M., Bin, A., Salam, A., Anjum, M.T., Kabir, A., & Kabir, A. (2024). A deep learning and machine learning approach to predict neonatal death in the context of São Paulo. International Journal of...

    2024

  52. [60]

    R., Hosain, M

    Abir, M. R., Hosain, M. T., Abdullah -Al-Jubair, M., & Mridha, M. F. (2024). IMVB7t: A Multi -Modal Model for Food Preferences based on Artificially Produced Traits. arXiv preprint arXiv:2412.16807

    2024 arXiv

  53. [225]

    doi:10.1109/TAC.2024.123456

    2024

  54. [420]

    doi:10.1145/192161.192272

  55. [560]

    doi:10.3389/fpsyg.2019.00560

    2019

  56. [1994]

    In Proceedings of the 21st Annual Conference on Computer Graphics and Interactive Techniques

    Animated Conversation: Rule -based Generation of Facial Expression, Gesture Spoken Intonation for Multiple Conversational Agents. In Proceedings of the 21st Annual Conference on Computer Graphics and Interactive Techniques. 413–

  57. [2007]

    Journal of management information systems 24, 3 (2007), 45 –77

    A design science research methodology for information systems research. Journal of management information systems 24, 3 (2007), 45 –77

    2007

  58. [2023]

    arXiv preprint arXiv:2307.06625 (2023)

    Automated Deception Detection from Videos: Using End -to-End Learning Based High -Level Features and Classification Approaches. arXiv preprint arXiv:2307.06625 (2023). https://arxiv.org/abs/2307.06625

    2023 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.