AI and Emotions: opportunities and challenges (Virtual Talk)
Talkat Understanding Higher-Level Intelligence from AI, Psychology, and Neuroscience Perspectives · 2024-06-24..06-28·
summary
Project started in 2016–17 builds a 23,000‑image corpus with 34 annotators and shows context plus body improves recognition over face alone; models use two‑stream and video networks, class activation maps, and find above‑chance but low absolute accuracy, with annotation distributions stabilizing after ~20 raters. AffectiveROAD experiments use AffectiveROAD: 9 drivers, 55 minutes and three stress levels, where video models beat single‑frame and multitask models outperform single‑task for predicting individual annotators; a Hyundai‑built kids’ assistive car is piloted in a Barcelona hospital.
AI-written summary · about
chapters (9)
- 0:00Emotion recognition in computer vision
- 7:57Early emotion recognition lessons
- 9:21Emotion perception in driving
- 16:30Assistive driving for children
- 18:19Cultural differences in emotion
- 22:58Ground truths in emotion modeling
- 25:53Social robotics and mental health use cases
- 26:57Open questions on AI emotions
- 28:28Q&A
related talks
14mAI Beyond Knowing - Explore Emotion AI with Neura
41mAffective Computing – bringing humans and machines closer through emotions
04mIncreasing business results through AI in the entertainment industry
01h 12mHumanizing AI – What It Means and Why It Matters
44mThe age of applied AI
21mAI Isn't Just Chat
06mCES Interview
31mAI & UX: Designing the Future of Human Interaction