Automated Face Recognition Systems as Adaptive Technology for Developmental Prosopagnosia
Face recognition plays a crucial role in social interactions; however, individuals with developmental prosopagnosia (DP) face significant challenges despite having normal intelligence and vision. DP affects approximately 2-4% of the population and can lead to social anxiety, embarrassment, and avoidance of social situations. Currently, there are no effective interventions for DP, and existing treatments have shown limited success.
This research will explore the potential of automated face recognition (AFR) algorithms (e.g., FaceNet), as an intervention for individuals with DP. This research has four studies that will collect self-reported preferences from DP participants regarding using AFR (Study 1), followed by modifying the algorithm to enable manual tagging of familiar faces. The performance of the algorithm will be assessed using a familiar face task, in both neurotypical (Study 2) and DP (Study 3) participants. Finally, the practical efficacy of the AFR algorithm as an intervention will be tested in a staged real-world scenario (Study 4). Data from the studies employing the AFR algorithm will be analysed using area under the receiver operating characteristic curve (AUC) scores, providing a comparative measure of AFR algorithm accuracy against neurotypical, DP, and DP using AFR algorithm performance.
The goal is to adapt an AFR algorithm for use in an app or other smart technology, as an intervention for people with DP. This type of intervention could be life-changing for individuals with DP and aligns well with the Transformative Technologies for Society challenge as it holds promise for improving accessibility for individuals with face recognition deficits.