Enhancing Customer Experiences with Emotional AI,
summary
Affectiva describes a concrete system: a deep‑learning computer‑vision pipeline that reads laptop or phone cameras in real time, maps facial micro‑movements to emotion, and returns moment‑by‑moment probabilities via cloud APIs or on‑device SDKs. Training uses human‑labeled video from 75 countries (uploader cites 5.5 million emotion samples, transcript adds ~2 billion frames), active learning, CNNs plus temporal models, and detects about 20 facial expressions → 8 emotional states. Commercial uses: ad testing (1,000‑person studies), driver attention and hiring video ranking; company refuses surveillance deals.
AI-written summary · about
chapters (13)
- 0:00Intro
- 4:08How did you get started in this area
- 6:41Tell us about Affectiva
- 10:13How does Affectiva work
- 12:53How do you convince the skeptic
- 15:21The circumplex model of emotions
- 16:17Measuring valence and arousal
- 17:37Hiring companies
- 18:43Deep learning approach
- 20:24Collecting data
- 21:54Data pipeline
- 25:42Platforms
- 30:00Wrap Up
related talks
50mAffective Computing
49mAffective Computing – Bringing Humans and Machines Closer Through Emotions
11mAmbient Intelligence and Emotionally Intelligent Devices
26mAffective Computing - What is it and why should I care?
46mAI That Reads Emotions: What It Means for Data Privacy, Consent, Your Teams
18mCommercial applications of AI
39mAI + JavaScript Rocks
52mPhysical AI: Crafting Resilient Systems with Emotional Intelligence