Emotion AI: A New Frontier Boisy Pitre Emotion AI Evangelist Affectiva @boisypitre @affectiva
Emotions influence every aspect of our lives @affectiva
From our health and wellbeing @affectiva
To what we purchase @affectiva
How we learn @affectiva
To the decisions we make @affectiva
And how we connect with each other @affectiva
We are surrounded by hyper-intelligent technologies with massive cognitive and autonomous capabilities @affectiva
All these systems have high “I.Q” but no EQ (Emotional Intelligence) @affectiva
Yet our interactions with technology are rapidly becoming… more conversational @affectiva
More relational @affectiva
What if our cars knew if we were getting angry and could make our brakes more responsive ? @affectiva
What if doctors could objectively measure how you are feeling the way they measure our other vital signs ? @affectiva
What if social robots, could actually be “social” ? @affectiva
What if every device in the internet of things had an emotion chip and could make personalized suggestions based on your mood? Your smart refrigerator could identify how you felt and suggest a food that matched your emotion. @affectiva
Our vision is to humanize technology with Emotion Intelligence by enabling machines to be emotion-aware and by allowing businesses to get emotion analytics. This is Emotion AI. @affectiva
Emotion AI How it works @affectiva
Our faces tell stories. @affectiva
SHAME PASSION PAIN LOVE CALM ANGER JOY GUILT @affectiva
What does emotion data look like? @affectiva
Telling the difference between… a Smile a Smirk @affectiva
Emotion Analysis in the Cloud @affectiva
Limitations of a Cloud-based API @affectiva
The case for a Real-Time API @affectiva
@affectiva
@affectiva
@affectiva
@affectiva
Just two seconds of… VIDEO AUDIO 496MB 32KB 1920x1080 32-bit frames at 30 16KHz 8-bit mono audio frames per second @affectiva
World’s largest emotion repository 75 countries, 5.3M faces analyzed, 2B facial frames Top Countries for Emotion Data UNITED KINGDOM 265K CHINA 502K JAPAN GERMANY 52K 126K USA 956K VIETNAM MEXICO 131K 117K PHILIPINES INDIA 133K BRAZIL 1,065K 176K THAILAND 162K INDONESIA 284K @affectiva
Guarding against biases Ensuring ethnicity, gender and age diversity in our training data Other African American South Asian Unknown Male Female South Asian Caucasian Hispanic @affectiva
What have we learned from the data ? @affectiva
Women more expressive than men! Smile duration (s) 60% 50% 45% 30% 40% Smile intensity 30% 30% 20% 15% 10% Females express more 0% than men 0 5 10 15 20 25 30 Man Woman @affectiva
Women smiled 32% more than men Women smiled more than men Women smiled in just over a quarter of the facial and their smiles were longer video responses, smiling in duration. 32% more than men. @affectiva
Men showed more “ brow furrowing ” compared to women Brow furrows from women were Women furrowed significantly shorter in duration in only 20% of the videos , and they actually furrow than those from men. 12% less than men @affectiva
Gender differences vary by culture In the US women smile 40% more than men, in France it’s 25%, but in the UK it’s the same 150% 125% 100% 75% 50% UNITED STATES FRANCE UNITED KINGDOM @affectiva
Older people more emotive +25% than younger people more 50 years below 50+ years @affectiva
In the US, pet care and baby care ads elicit the most enjoyment! Enjoyment 9 Pet Care Baby Care 8 Dairy Entertainment 7 Snacks Ice cream 6 Telecom Chocolate Enjoyment 5 Apparel Auto 4 Restaurant Services 3 Prescription 2 Fuel 1 0 1 2 3 4 5 6 7 8 9 10 @affectiva
In Canada, its the Cereal ads that elicit the most enjoyment! Enjoyment Cereal 9 8 Finance OTC Beverage 7 6 Entertainment Enjoyment 5 Chocolate 4 Spreads 3 Candy 2 Telecom 1 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 @affectiva
Emotion AI How it is used @affectiva
How Emotion AI is used Insights & Analytics Real Time Interaction Understand how people engage Apps and digital experiences adapt emotionally in the digital world to our emotions in real time @affectiva
Gaming Nevermind Psychological thriller game becomes more challenging when sensing player distress @affectiva
Video Recruitment Top performers are more expressive 180 They smile more, but its not just Bottom 20% Top 80% 160 that. They show a wider range of emotion compared to bottom 140 performers – maybe they are more 120 Percent Change authentic? 100 80 Data pulled from 13,356 60 interviews that also had 40 performance model scores. 20 0 Anger (range) Fear (range) Sadness (max) Attention (range) Joy (range) Valence (max) Smile (range) Surpise (range) Contempt (range) Disgust (max) Smirk (range) @affectiva
Health and Wellbeing Brain Power: helping individuals on the autism spectrum World’s First Augmented Reality Smart-Glass-System empowers children and adults with autism to teach themselves crucial social and cognitive skills using Affectiva emotion recognition tech. @affectiva
We’ve spent the past 50 years interacting with computers with their language. The time has come where computers can interact with us using our language. @affectiva
Combining Face and Voice Metrics @affectiva
Facing the Road Ahead @affectiva
Learn more at affectiva.com @affectiva Boisy Pitre @boisypitre @affectiva
Recommend
More recommend
