Deception detection in videos

Abstract

Deception detection is a growing area of research with applications in security, forensic investigations, and misinformation analysis. Traditional methods such as polygraphs and human behavioral analysis are limited by subjectivity and susceptibility to countermeasures. The creation of automatic deception detection classifiers is critical in overcoming all these limitations. They analyze speech, text, and facial expressions to identify deceptive cues, leveraging machine learning, and multi-modal data fusion to enhance accuracy and reliability.

This project focuses on refining feature extraction and improving the model framework while tackling challenges such as dataset bias and real-world applicability. The report will provide an overview of the models used for deception detection in previous studies. The purpose of this paper is to provide an in-depth analysis of modifications to the framework, explaining how each component collectively contributes to model refinement. A proposed automated deception detection classifier will be implemented, trained, and evaluated against previous classifiers.