作者: Víctor Ponce-López , Baiyu Chen , Marc Oliu , Ciprian Corneanu , Albert Clapés
DOI: 10.1007/978-3-319-49409-8_32
关键词:
摘要: This paper summarizes the ChaLearn Looking at People 2016 First Impressions challenge data and results obtained by teams in first round of competition. The goal competition was to automatically evaluate five “apparent” personality traits (the so-called “Big Five”) from videos subjects speaking front a camera, using human judgment. In this edition challenge, novel set consisting 10,000 shorts clips YouTube has been made publicly available. ground truth for workers Amazon Mechanical Turk (AMT). To alleviate calibration problems between workers, we used pairwise comparisons videos, variable levels were reconstructed fitting Bradley-Terry-Luce model with maximum likelihood. CodaLab open source platform submission predictions scoring. attracted, over period 2 months, 84 participants who are grouped several teams. Nine entered final phase. Despite difficulty task, great advances challenge.