• Priya R.Ghongade
• Prof.Vijaykumar M.P

Automated deception detection seeks to infer the veracity of human communication from observable behavioural signals rather than from invasive physiological instrumentation. Conventional polygraphy, while historically dominant, is confounded by its reliance on autonomic arousal, its intrusiveness, and its contested scientific validity. The emergence of deep learning has enabled non-invasive analysis of facial expressions and micro-expressions using convolutional neural networks (CNNs), motivating sustained research interest in visual, facial-video-based deception analysis. This review synthesises the state of the art in CNN-based visual and facial deception detection, examining spatial feature extraction, temporal extensions such as CNN-LSTM architectures for facial expression sequences, and the wider behavioural-science literature on facial action units, micro-expressions, and eye gaze/blink dynamics that underpins this line of research. Widely used facial-emotion resources such as FER2013 and CK+ are discussed and explicitly distinguished from purpose-built deception corpora with a visual/facial channel, including the Real-Life Trial dataset. The review develops a taxonomy of CNN-based visual/facial approaches, synthesises findings from the literature into a comparative table, and identifies persistent research gaps concerning dataset scarcity, cross-domain generalisation, the loss of temporal information in static-frame CNN analysis, explainability, demographic and cultural bias, and real-time deployment. A recurring theme is that facial and behavioural arousal are correlates of stress or cognitive load rather than direct markers of falsehood, and this distinction is treated as a central methodological caution throughout. A conceptual, literature-grounded CNN-based visual pipeline is proposed as a direction for future research rather than as a validated system. The review concludes that CNN-based facial analysis is best positioned as a probabilistic, explainable, human-in-the-loop decision-support approach for deception-related behavioural cues, not as a definitive lie detector.

Deception detection; convolutional neural networks; facial expression recognition; facial micro-expressions; Facial Action Units; visual behavioural analysis; deep learning; explainable artificial intelligence

"CNN-Based Visual and Facial Deception Detection: A Structured Review of Facial Expression, Micro-Expression, and Behaviour Cue Analysis", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 9, page no.c481-c489, September-2026, Available :http://www.jetir.org/papers/JETIR2609251.pdf

"CNN-Based Visual and Facial Deception Detection: A Structured Review of Facial Expression, Micro-Expression, and Behaviour Cue Analysis", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 9, page no. ppc481-c489, September-2026, Available at : http://www.jetir.org/papers/JETIR2609251.pdf