113% ROI on Large Enterprise Crowdsourcing Programs - Forrester TEIā„¢ Get the Study ×

The Contextual is King – Announcing The Face Emotion Analysis Marathon Match

By wwwtc In Uncategorized

Posted December 10th, 2014

“Let’s eat, grandpa.” or “Let’s eat grandpa.”

The comma. It is the vital pause between words; the fulcrum for meaning when you say, “Let’s eat, grandpa.” That important contextual pause between words can result in an audience who feels happiness at the memory of dinner with their grandfather, or an audience overcome with shock and disgust.

Academic research is clear that much of human communication happens through non-verbal cues. As much as sound and words, body language communicates meaning. Scientists assert that humans can distinguish at least six basic emotions: happiness, sadness, fear, anger, surprise, and disgust. In fact, the vast majority of person-to-person communication across global cultures is … unspoken.

But what of computers and software? Can they too be made to interpret these 6 non-verbal human emotions? And, perhaps more importantly, when it comes to Face Recognition software, what good is an algorithm if it can’t measure the contextual human data being communicated?

Announcing a new Topcoder Data Science Challenge:

The Face Emotion Analysis Marathon Match >>

In this Topcoder Marathon Match, HP is challenging the Topcoder Data Science community to develop new algorithms to detect human emotions in photos of facial expressions. HP recently released a Face Detection API that can analyze an image to find faces, and estimate the ages of people in the image. In this Topcoder data science challenge, we are looking for Topcoder community members to create an algorithm that recognizes the human emotions present in a database of images. Emotions like: anger, anxiety, confidence, happiness, sadness, and surprise.

What is HP IDOL OnDemand? It is growing list of 50+ REST APIs that are drop-dead-simple to use for unstructured data … the messy, but vital, human data that does not fit easily into a table or database.

Below is the prize money that will be awarded for the top 5 performing algorithms – and you can find the contest details for this Topcoder contest here.

  • 1st place: $5,000
  • 2nd place: $2,000
  • 3rd place: $1,500
  • 4th place: $1,000
  • 5th place: $500

Your winning algorithms will help accurately detect the all important human contextual message. So – tell us, is the person in this picture: Happy? Surprised? Sad? or Amazed?

Good luck Topcoder Community members. We are excited to see what you come up with!

And remember: Human data without context is a statistic, a metric. Contextual human data, well … that is where meaning can be found.

Time for dinner, Grandpa.