Seeing Faces in Things: A Model and Dataset for Pareidolia

Explore the fascinating phenomenon of face pareidolia in this 51-minute lecture by Bill Freeman from MIT. Delve into the human visual system's tendency to detect faces in random stimuli, such as coffee stains or clouds. Learn about the "Faces in Things" dataset, consisting of 5,000 annotated images showcasing pareidolic faces. Examine the behavioral gap between humans and state-of-the-art neural networks in detecting these illusory faces. Discover how the human ability to recognize both human and animal faces contributes to this discrepancy. Investigate a proposed statistical model for general object pareidolia and its predictions regarding image conditions most likely to induce this effect. Gain insights into lower-level intelligence from AI, psychology, and neuroscience perspectives through this comprehensive exploration of face pareidolia.

Seeing Faces in Things: A Model and Dataset for Pareidolia