Pith. sign in

REVIEW 4 major objections 5 minor 36 references

A detector that concatenates LipNet, DeepSpeech2, and ResNet18 features and classifies with Random Forest reaches 94 percent accuracy on FakeAVCeleb.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 10:57 UTC pith:6GGN726H

load-bearing objection A plausible feature-concatenation pipeline for deepfake detection, but the headline 94% accuracy is unverifiable because the paper never states whether the 70/15/15 split is at the video level or the instance level. the 4 major comments →

arxiv 2607.20579 v1 pith:6GGN726H submitted 2026-07-22 cs.CR

Deepfake News Detection: A Multimodal Framework Integrating LipNet, DeepSpeech and ResNET for Enhanced Audio-Visual Analysis

Ameena Khan , Muhammad Ahsan Aziz , Muhammad Junaid Asif , Naeem Akhter , Rana Fayyaz Ahmad This is my paper
classification cs.CR
keywords deepfake detectionmultimodal audio-visuallip-sync analysisspeech-to-text featuresRandom ForestFakeAVCelebsliding-window augmentationdeepfake news
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper tries to establish that fake news videos can be caught by joining features from lip motion, speech content, and facial appearance and feeding the combined vector to standard classifiers. On the FakeAVCeleb dataset, the proposed Random Forest pipeline reaches 94% accuracy when trained on augmented audio and lip-sync features, beating a published multimodal ensemble baseline that scores 89%. The paper argues that the decisive evidence is in audio and lip-sync coherence: adding face appearance features slightly lowers accuracy to 93%, because current generators create visually convincing faces. The same pipeline drops to 52% on the DFDC dataset, which the authors attribute to that dataset lacking audio-specific labels. If the result holds, a relatively simple and efficient audio-first detector could screen manipulated news content near real time.

Core claim

The central claim is that concatenating transfer-learned features from LipNet and DeepSpeech2 and classifying them with a Random Forest yields 94% accuracy on FakeAVCeleb, outperforming a state-of-the-art multimodal ensemble baseline at 89%. The authors attribute the largest gain to sliding-window augmentation over one-second segments, which raises audio-only accuracy from 86% to 94% and F1 from 0.8712 to 0.9424. They also claim that visual features have reached diminishing returns: adding ResNet18 face features does not improve accuracy, settling at 93.34%, because modern fake faces are visually indistinguishable. The paper presents this as evidence that deepfake news detection should empha

What carries the argument

The mechanism is feature concatenation across three extractors. LipNet encodes cropped mouth regions into lip-motion features; DeepSpeech2 converts one-second Mel spectrograms into speech-content features; and BlazeFace detects and crops faces that ResNet18 encodes into appearance features. The concatenated vector is classified by Random Forest, which the experiments show handles the high-dimensional, non-linear feature space better than MLP, SVM, or LSTM. Sliding-window augmentation over one-second segments is the component credited with the largest accuracy improvement, expanding the minority class and exposing classifiers to varied acoustic conditions.

Load-bearing premise

The 94% accuracy assumes that no one-second segment from the same source video appears in both the training and test sets; the paper does not state whether the split is at the video level or the segment level.

What would settle it

Rerun the Random Forest training and testing after splitting FakeAVCeleb at the video level, so all one-second segments from a source video stay in one partition. If accuracy falls well below 94%, the reported result was inflated by segment leakage rather than genuine generalization.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the 94% figure is correct, audio and lip-sync features alone are sufficient to outperform a multimodal ensemble baseline on FakeAVCeleb, making facial appearance optional for this benchmark.
  • The 86 to 94 percentage point gain shows that one-second sliding-window augmentation is an effective counter to class imbalance in synthetic media datasets.
  • The 93% result with visual features implies that as face generators improve, detectors should downweight facial appearance and rely on lip-sync and audio coherence.
  • The 52% DFDC result implies that datasets without audio-specific labels are unsuitable for training or evaluating audio-visual deepfake detectors.
  • A Random Forest on a concatenated feature vector is computationally light enough to support near-real-time screening of news video.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the same 94% accuracy survives a strict video-level split, audio-first detection could serve as a cheap first-pass filter in news verification, though cross-dataset accuracy would likely drop because FakeAVCeleb is a single benchmark.
  • The claim that faces are unhelpful may be specific to FakeAVCeleb's generation process; a direct test is to rerun the pipeline on newer deepfake datasets and watch whether the weight of visual features shifts.
  • Segment-level predictions could be used to localize manipulated portions of a video, turning the classifier into a temporal forensics tool rather than a whole-video labeler.
  • A stronger evaluation for real-world deployment would be per-video majority voting over one-second predictions under a video-level split.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes a deepfake-news detection framework that extracts features from lip movements (LipNet), speech content (DeepSpeech2), and facial appearance (BlazeFace + ResNet18), concatenates them, and classifies the resulting vector with Random Forest, MLP, LSTM, and several other learners. Experiments are reported on FakeAVCeleb, with the claim that augmented audio features give 94% accuracy (Random Forest, Table III), outperforming a state-of-the-art multimodal ensemble baseline [25]. A second dataset, DFDC, is mentioned but only a 52% accuracy is reported in passing. The manuscript stresses a 'semantic-based feature analysis' and efficiency-oriented selection of informative sub-networks, but no such analysis is actually presented.

Significance. If the 94% accuracy figure were valid and reproducible, the proposed pipeline would be a simple, efficient, audio-centric detector for deepfake news videos, and the comparison against the ensemble baseline in Table III would be of interest to the community. The paper's strength is the systematic comparison of many classifiers on the official FakeAVCeleb categories and its clear tabular reporting. However, the central claim currently rests on two unresolved protocol issues: the split granularity for the sliding-window augmentation is not specified, creating a likely data-leakage path, and the audio-only configuration is selected after observing the audio-visual result despite the paper's promise of a principled feature-selection analysis. Without code, split indices, or a corrected experiment, the headline number cannot be taken as evidence for the framework's claimed performance.

major comments (4)
  1. [§IV-B and §IV-C] The split granularity is not specified. The paper says each video is preprocessed into 'five or more real instances' and augmentation is done 'by sliding a window over one-second segments' (§IV-B), and then 'Seventy percent of the data is used for training, fifteen percent for validation and fifteen percent for testing' (§IV-C). Since 500 real videos become 50,000 real instances (about 100 segments per source video), an instance-level split would place nearly identical, overlapping segments from the same video in both training and test sets, letting Random Forest memorize source-video-specific cues. This would directly inflate the 94.36% accuracy in Table III. The paper must state whether the split was performed at the video level, and if so, provide the split indices or a video-grouped evaluation. Without this, the central claim is not interpretable.
  2. [§V-D and §III] The audio-only configuration is selected post hoc, contradicting the paper's stated methodology. Section III promises that 'the covariance between them is examined' and that 'only the sub-networks that show the greatest effectiveness are kept.' No covariance analysis, ablation table, or feature-selection results appear in §V. Instead, §V-D reports that the audio-visual Random Forest accuracy is 0.9334 and then states this 'motivates the design choice of emphasizing audio and lip-sync coherence.' This is circular: the model is evaluated on the test set, and the best-performing configuration is then described as the intended design. The authors should provide the promised semantic feature analysis, or clearly state that the audio-only configuration was chosen after observing validation performance, and re-evaluate accordingly.
  3. [§V-A and §I] The manuscript claims robust generalization across diverse datasets, but its only DFDC result is a 52% accuracy, reported without details (§V-A). The introduction claims the goal is 'robust generalization across diverse datasets,' and the abstract/conclusion claim 'robustness and practical potential.' Near-chance performance on DFDC directly contradicts these generalization claims. If DFDC is retained in the narrative, the authors should report the full metric set (precision, recall, F1) for the selected configuration and either temper the generalization claims or explain why DFDC is out of scope. As written, the gap between the claimed robustness and the actual results is too large to ignore.
  4. [§V-D and Fig. 10] The assertion that 'the updated dataset contains synthetic faces that are visually indistinguishable from real ones' is used to justify the decision to drop visual features. This is an empirical claim with no supporting human evaluation or quantitative analysis; the single example in Fig. 10 is anecdotal. The lack of improvement from adding ResNet18 features could equally be due to poor face alignment, feature extraction settings, or the leakage issue above. If this explanation is retained, it needs evidence; otherwise the claim should be removed or qualified.
minor comments (5)
  1. [§IV-B] The augmentation description is inconsistent with the stated class balance: FakeAVCeleb has ~500 real and ~19,500 fake videos, but after preprocessing the paper reports 50,000 real and 100,000 fake instances. The augmentation multiplier (100× for real, ~5.1× for fake) is not explained, and the resulting 1:2 ratio is still imbalanced. Please clarify the exact augmentation procedure and whether the same procedure was applied to both classes.
  2. [Fig. 3 and §III-C] Figure 3 lists eight classifiers (Random Forest, MLP, AdaBoost, Decision Trees, SVM, K-Nearest Neighbor, Naïve Bayes, LSTM), but §III-C describes only Random Forest, MLP, and LSTM. The remaining classifiers are never described in the methodology, even though they appear in the results tables. Either add brief descriptions or remove them from the figure.
  3. [§III-D] The 'Fine-Tuning' subsection is vague: it says weights are reused from 'existing detection models' and only the final classification layer is specialized, but no details are given about which layers were frozen, which were fine-tuned, or the optimization schedule. Please provide concrete fine-tuning settings or remove this subsection.
  4. [References [26]-[29]] Some references are non-archival or inappropriate: DeepSpeech2 is cited to a blog post, BlazeFace to a personal GitHub repository, and ResNet18 is cited to an Alzheimer's-disease classification paper rather than the original ResNet paper. Please replace with the primary sources.
  5. [Tables III and V] The comparison with baseline [25] in Table III reports a single accuracy (0.8943), but no information is given about whether this baseline was re-implemented on the same split and with the same augmentation. Without this, the 'surpassing the state-of-the-art' claim in Table V is not fully supported. Please describe the baseline evaluation protocol.

Circularity Check

1 steps flagged

Partial circularity: the headline 94% audio-only Random Forest result is the post-hoc best configuration selected from the same test-accuracy tables, then presented as the proposed method's performance.

specific steps
  1. fitted input called prediction [§I and §III (semantic feature analysis); §V-C/§V-D/§V-E (Tables III-V and discussion)]
    "a semantic-based feature analysis technique is introduced in which each sub-network is independently assessed and only the most effective outputs are retained ... Random Forest attains 94 percent accuracy, surpassing the state-of-the-art ensemble multimodal baseline [25] ... This finding motivates the design choice of emphasizing audio and lip-sync coherence rather than relying on raw facial appearance."

    The configuration that becomes 'the proposed approach' (Random Forest on augmented audio-only features) is chosen after inspecting the test-accuracy tables: audio-only beats audio-visual (Table III vs Table IV) and Random Forest beats every other classifier. The stated selection rule—'only the most effective outputs are retained'—uses the same accuracy metric as the headline claim, so the 94% figure is the maximum of the tested grid, not an independent prediction of a pre-specified model. No covariance/semantic analysis is shown, so the choice is empirically forced by the reported numbers rather than by the methodology.

full rationale

The paper is mostly a standard empirical evaluation with external anchors: the FakeAVCeleb dataset [30] and the ensemble baseline [25] are external, and the low DFDC accuracy is reported as a limitation rather than hidden. The self-citations [32]-[36] are for standard metric equations and are not load-bearing. However, the central 94% claim is partially circular because the paper presents a 'semantic feature analysis' that is never specified, and the observable basis for the audio-only Random Forest configuration is its top accuracy in the same tables used for the headline. A separate non-circular validity threat is that sliding-window augmentation ('sliding a window over one-second segments', §IV-B) could straddle the 70/15/15 split because split granularity is not stated; that is a leakage/correctness risk, not a definitional reduction, so it does not by itself raise the circularity score beyond the model-selection issue.

Axiom & Free-Parameter Ledger

5 free parameters · 5 axioms · 0 invented entities

The reported 94% depends on several hand-set choices (segment length, augmentation targets, fixed training budget) and on unstated modeling assumptions about window labels and split independence. None of these parameters are fitted by a principled criterion, and no sensitivity analysis is provided.

free parameters (5)
  • audio segment length = 1 second
    Controls the window of spectrograms fed to LipNet/DeepSpeech2 and the augmentation stride; chosen by hand with no sensitivity analysis (§IV-B, Table I).
  • augmentation target counts = 50,000 real / 100,000 fake instances
    Sliding-window augmentation expands 500 real videos to 50,000 real instances (100x) and ~19,500 fake videos to 100,000 fakes; the expansion factors and class balance affect all reported accuracies and are not justified (§IV-B).
  • training steps = 100
    Fixed number of training steps for fine-tuning; no convergence analysis or early stopping (§IV-C, Table I).
  • learning rate = 0.01
    Gradient-descent learning rate for MLP fine-tuning; no sensitivity analysis (§IV-C, Table I).
  • LSTM sequence length = 1
    The paper sets sequence length to one, converting LSTM into a feedforward network (§III-C); this hand-set choice undermines the use of LSTM and discards temporal modeling.
axioms (5)
  • domain assumption Sliding-window segments are treated as independent samples for the 70/15/15 split
    Required for the reported accuracy to be unbiased; no video-level split is described (§IV-B), so segments from the same source video may appear in both training and test sets.
  • domain assumption Binary real/fake labels on FakeAVCeleb are valid for 1-second augmented windows
    The paper labels every 1-second segment as real or fake, but manipulation boundaries may fall inside a window, and the dataset's four fine-grained categories are collapsed to binary (§IV-A/B).
  • ad hoc to paper Synthetic faces are visually indistinguishable from real faces
    Invoked to explain why the audio+video configuration (93%) underperforms audio-only (94%) and to justify de-emphasizing visual features (§V-D, Fig. 10). This is a post-hoc explanation, not a measured property.
  • domain assumption Video features have no temporal dependencies between frames
    Used to set LSTM sequence length to 1 (§III-C), effectively discarding the recurrent mechanism; contradicts temporal structure used by most deepfake detectors.
  • domain assumption DFDC labels are inadequate for audio-visual fine-grained analysis
    Used to dismiss the 52% DFDC result (§V-A), though DFDC is a standard benchmark for frame-level video forgery.

pith-pipeline@v1.3.0-alltime-deepseek · 10833 in / 15533 out tokens · 193118 ms · 2026-08-01T10:57:29.471711+00:00 · methodology

0 comments
read the original abstract

Deepfake news refers to AI-generated (or AI ma-nipulated) multimedia content intentionally generated to deceive audiences by manipulating the facial expressions, or speech while maintaining the realistic appearance. The rapid progress of generative AI has made the synthesis of highly realistic fake videos and cloned voices widely accessible, posing a serious threat to the authenticity of digital news media. This paper presents a multi-modal framework that discerns the authenticity of video content by jointly exploiting audio and visual cues, thereby addressing the challenge of detecting the deepfake videos. We proposed a framework that involves features extraction from lip movements, audio content and video frames. Lip movements and speech content are encoded using the LipNet and DeepSpeech2 models, while facial features are extracted by leveraging the use of BlazeFace and represented with ResNet18. The extracted feature vectors are concatenated into a holistic video representation and classified with an ensemble of machine learning and deep learning models, including Random Forest (RF), Multi-layer Perceptron (MLP) and Long Short-Term Memory (LSTM) networks. Exten-sive experiments performed on the FakeAVCeleb dataset shows that the proposed approach attains an accuracy of 94% using augmented audio features, outperforming a state-of-the-art multi-modal ensemble baseline. The results confirm the robustness and practical potential of the proposed framework for deepfake news detection.

Figures

Figures reproduced from arXiv: 2607.20579 by Ameena Khan, Muhammad Ahsan Aziz, Muhammad Junaid Asif, Naeem Akhter, Rana Fayyaz Ahmad.

Figure 1
Figure 1. Figure 1: Overview of the deepfake generation process based on an autoencoder [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed multimodal feature extraction [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 5
Figure 5. Figure 5 [PITH_FULL_IMAGE:figures/full_fig_p003_5.png] view at source ↗
Figure 3
Figure 3. Figure 3: Classification stage of the proposed framework. The concatenated feature vector is fed to an ensemble of classifiers including Random Forest, MLP, [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Architecture of the LipNet feature extraction module. Sequential [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: DeepSpeech2 model: invariant convolution layers, batch normalisation [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: BlazeFace model output: a bounding box and six facial landmarks [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Step-wise augmentation using a sliding sample window over one [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Accuracy of the classification models for audio features [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 10
Figure 10. Figure 10: Example of a synthetic (fake) person that is visually indistinguishable [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Accuracy of the classification models for audio–visual features [PITH_FULL_IMAGE:figures/full_fig_p008_11.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 6 linked inside Pith

  1. [1]

    Generative adversarial networks,

    I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative adversarial networks,” arXiv:1406.2661, 2014

    Pith/arXiv arXiv 2014

  2. [2]

    Jordan peele’s simulated obama psa is a double-edged warning against fake news,

    A. Romano, “Jordan peele’s simulated obama psa is a double-edged warning against fake news,” Vox, 2018

    2018

  3. [3]

    Chinese deepfake app zao goes viral, faces immediate criticism over user data and security policy,

    J. Damiani, “Chinese deepfake app zao goes viral, faces immediate criticism over user data and security policy,” Forbes, 2021

    2021

  4. [4]

    An AI program will soon be here to help your deepfake dancing – just don’t call it deepfake,

    P. Farquhar, “An AI program will soon be here to help your deepfake dancing – just don’t call it deepfake,” Business Insider, 2018

    2018

  5. [5]

    Syn- thesizing obama: Learning lip sync from audio,

    S. Suwajanakorn, S. M. Seitz, and I. Kemelmacher-Shlizerman, “Syn- thesizing obama: Learning lip sync from audio,” ACM Trans. Graph., vol. 36, 2017

    2017

  6. [6]

    Xception: Deep learning with depthwise separable convo- lutions,

    F. Chollet, “Xception: Deep learning with depthwise separable convo- lutions,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2017

    2017

  7. [7]

    MesoNet: A compact facial video forgery detection network,

    D. Afchar, V. Nozick, J. Yamagishi, and I. Echizen, “MesoNet: A compact facial video forgery detection network,” in Proc. IEEE Int. Workshop Inf. Forensics Secur. (WIFS), 2018

    2018

  8. [8]

    Deepfake video detection using recurrent neural networks,

    D. Gu¨era and E. J. Delp, “Deepfake video detection using recurrent neural networks,” in Proc. IEEE Int. Conf. Adv. Video Signal Based Surveillance (AVSS), 2018

    2018

  9. [9]

    FakeCatcher: Detection of synthetic portrait videos using biological signals,

    U. Ciftci and I. Demir, “FakeCatcher: Detection of synthetic portrait videos using biological signals,” arXiv:1901.02212, 2019

    Pith/arXiv arXiv 1901

  10. [10]

    In ictu oculi: Exposing AI created fake videos by detecting eye blinking,

    Y. Li, M.-C. Chang, and S. Lyu, “In ictu oculi: Exposing AI created fake videos by detecting eye blinking,” in Proc. IEEE Int. Workshop Inf. Forensics Secur. (WIFS), 2018

    2018

  11. [11]

    Exposing deepfake videos by detecting face warping artifacts,

    Y. Li and S. Lyu, “Exposing deepfake videos by detecting face warping artifacts,” arXiv:1811.00656, 2018

    Pith/arXiv arXiv 2018

  12. [12]

    Face x-ray for more general face forgery detection,

    L. Li, J. Bao, T. Zhang, H. Yang, D. Chen, F. Wen, and B. Guo, “Face x-ray for more general face forgery detection,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), 2020

    2020

  13. [13]

    Relative phase information for detecting human speech and spoofed speech,

    L. Wang, Y. Yoshida, Y. Kawakami, and S. Nakagawa, “Relative phase information for detecting human speech and spoofed speech,” in Proc. 16th Annu. Conf. Int. Speech Commun. Assoc. (INTERSPEECH), 2015

    2015

  14. [14]

    Combining evidences from mel cepstral, cochlear filter cepstral and instantaneous frequency features for detection of natural vs. spoofed speech,

    T. B. Patel and H. A. Patil, “Combining evidences from mel cepstral, cochlear filter cepstral and instantaneous frequency features for detection of natural vs. spoofed speech,” in Proc. 16th Annu. Conf. Int. Speech Commun. Assoc. (INTERSPEECH), 2015

    2015

  15. [15]

    Not made for each other–audio-visual dissonance-based deepfake detection and local- ization,

    K. Chugh, P. Gupta, A. Dhall, and R. Subramanian, “Not made for each other–audio-visual dissonance-based deepfake detection and local- ization,” in Proc. 28th ACM Int. Conf. Multimedia, 2020

    2020

  16. [16]

    Emotions don’t lie: An audio-visual deepfake detection method using affective cues,

    T. Mittal, U. Bhattacharya, R. Chandra, A. Bera, and D. Manocha, “Emotions don’t lie: An audio-visual deepfake detection method using affective cues,” in Proc. 28th ACM Int. Conf. Multimedia, 2020

    2020

  17. [17]

    Lip sync matters: A novel multimodal forgery detector,

    S. A. Shahzad, A. Hashmi, S. Khan, Y.-T. Peng, Y. Tsao, and H.-M. Wang, “Lip sync matters: A novel multimodal forgery detector,” in Proc. IEEE, 2022

    2022

  18. [18]

    A comparison of features for synthetic speech detection,

    M. Sahidullah, T. Kinnunen, and C. Hanilc¸i, “A comparison of features for synthetic speech detection,” in Proc. INTERSPEECH, 2015

    2015

  19. [19]

    Multimodaltrace: Deepfake detection using audiovisual representation learning,

    M. A. Raza and K. M. Malik, “Multimodaltrace: Deepfake detection using audiovisual representation learning,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), 2023

    2023

  20. [20]

    Deepfake video detection based on spatial, spectral, and temporal inconsistencies us- ing multimodal deep learning,

    J. K. Lewis, I. E. Toubal, H. Chen, V. Sandesera, M. Lomnitz, Z. Hampel-Arias, C. Prasad, and K. Palaniappan, “Deepfake video detection based on spatial, spectral, and temporal inconsistencies us- ing multimodal deep learning,” in Proc. IEEE Appl. Imagery Pattern Recognit. Workshop (AIPR), 2020

    2020

  21. [21]

    Emotions don’t lie: An audio-visual deepfake detection method using affective cues,

    T. Mittal, U. Bhattacharya, R. Chandra, A. Bera, and D. Manocha, “Emotions don’t lie: An audio-visual deepfake detection method using affective cues,” in Proc. ACM Int. Conf. Multimedia, 2020

    2020

  22. [22]

    Not made for each other–audio-visual dissonance-based deepfake detection and local- ization,

    K. Chugh, P. Gupta, A. Dhall, and R. Subramanian, “Not made for each other–audio-visual dissonance-based deepfake detection and local- ization,” in Proc. ACM Int. Conf. Multimedia, 2020

    2020

  23. [23]

    Evaluation of an audio- video multimodal deepfake dataset using unimodal and multimodal detectors,

    H. Khalid, M. Kim, S. Tariq, and S. S. Woo, “Evaluation of an audio- video multimodal deepfake dataset using unimodal and multimodal detectors,” in Proc. 1st Workshop Synthetic Multimedia-Audiovisual Deepfake Gener. Detection, 2021

    2021

  24. [24]

    AVFakeNet: A unified end-to-end dense swin transformer deep learning model for audio-visual deepfakes detection,

    H. Ilyas, A. Javed, and K. M. Malik, “AVFakeNet: A unified end-to-end dense swin transformer deep learning model for audio-visual deepfakes detection,” Appl. Soft Comput., 2023

    2023

  25. [25]

    Multimodal forgery detection using ensemble learning,

    A. Hashmi, S. A. Shahzad, W. Ahmad, C.-W. Lin, Y. Tsao, and H.-M. Wang, “Multimodal forgery detection using ensemble learning,” in Proc. IEEE, 2022

    2022

  26. [26]

    LipNet: End-to-end sentence-level lipreading,

    Y. M. Assael, B. Shillingford, S. Whiteson, and N. de Freitas, “LipNet: End-to-end sentence-level lipreading,” arXiv:1611.01599, 2016

    Pith/arXiv arXiv 2016

  27. [27]

    DeepSpeech2: A machine learning model for speech recognition,

    D. Cochard, “DeepSpeech2: A machine learning model for speech recognition,” axinc-ai, 2021

    2021

  28. [28]

    hollance/BlazeFace-PyTorch,

    M. Hollemans, “hollance/BlazeFace-PyTorch,” 2020

    2020

  29. [29]

    A deep learning approach for automated diagnosis and multi-class classification of alzheimer’s disease stages using resting- state fMRI and residual neural networks,

    F. Ramzan, M. U. G. Khan, A. Rehmat, S. Iqbal, T. Saba, A. Rehman, and Z. Mehmood, “A deep learning approach for automated diagnosis and multi-class classification of alzheimer’s disease stages using resting- state fMRI and residual neural networks,” J. Med. Syst., vol. 44, 2019

    2019

  30. [30]

    FakeAVCeleb: A novel audio-video multimodal deepfake dataset,

    H. Khalid, S. Tariq, and S. S. Woo, “FakeAVCeleb: A novel audio-video multimodal deepfake dataset,” arXiv:2108.05080, 2021

    Pith/arXiv arXiv 2021

  31. [31]

    The deepfake detection challenge (DFDC) preview dataset,

    B. Dolhansky, R. Howes, B. Pflaum, N. Baram, and C. Fer- rer, “The deepfake detection challenge (DFDC) preview dataset,” arXiv:1910.08854, 2019

    Pith/arXiv arXiv 1910

  32. [32]

    Crowd scene analysis using deep learning techniques,

    M. J. Asif, “Crowd scene analysis using deep learning techniques,” Ph.D. dissertation, University of Central Punjab (UCP) Lahore, Pakistan, 2024

    2024

  33. [33]

    Conv- lstm for real-time spatio-temporal analysis of crowd behavior in public spaces,

    M. J. Asif, S. Saqib, R. F. Ahmad, M. Asad, and S. T. H. Rizvi, “Conv- lstm for real-time spatio-temporal analysis of crowd behavior in public spaces,” in 4th International Conference on Computing, Communica- tions and Digital Systems (C-Code 2025). IEEE x Bahria University Islamabad, 2025

    2025

  34. [34]

    Exploring convolutional neural networks for rice grain classification: An explainable ai approach,

    M. J. Asif, H. Khan, R. Tehseen, S. T. H. Rizvi, M. Asad, S. Saqib, and R. F. Ahmad, “Exploring convolutional neural networks for rice grain classification: An explainable ai approach,” in 2025 International Conference on Frontiers of Information Technology (FIT). IEEE, 2025

    2025

  35. [35]

    Towards automated solar panel integrity: Hybrid deep feature extraction for advanced surface defect identification,

    M. J. Asif, M. S. Rafaqat, U. Nazakat, U. Khan, and R. F. Ahmad, “Towards automated solar panel integrity: Hybrid deep feature extraction for advanced surface defect identification,” 2026

    2026

  36. [36]

    Strategic customer segmentation: Harnessing machine learning for retaining satisfied cus- tomers,

    H. Khalid, S. Saqib, M. J. Asif, and D. A. Dewi, “Strategic customer segmentation: Harnessing machine learning for retaining satisfied cus- tomers,” Lahore Garrison University Research Journal of Computer Science and Information Technology, vol. 8, no. 2, 2024

    2024