DECEPTIcON: Bridging Gaps in In-the-Wild Deception Research

• Published in: IEEE transactions on affective computing pp. 1 - 13
• Main Authors: Bicer, Berat, Durmaz, Bahadlr, Aras, Serhat, Dibeklioglu, Hamdi
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: Accuracy; Affective computing; Annotations; classification metrics; cross-validation; fact-checking; Fake news; Feature extraction; multimodal analysis; Organizations; political deception; Standards organizations; transformer architectures; Transformers; Visualization

We present DECEPTIcON , a new large-scale dataset for automatic deception detection. It contains video clips from 100 public figures, mostly politicians, along with manually aligned text transcripts and extracted audio-visual features. Each video is labeled with one of six truth levels from the Poli...