• Use cases
  • How it works
  • Developers

Now is the time to make artificial intelligence
​less artificial and

​more human

Predict visual emotion using deep learning

Neon Open uses the brain's emotional response to images to inform behavioral experiments that can be crowd sourced to collect data to train convolutional deep neural networks. 
Brain response
Picture
Crowd-sourced Data
Picture
Deep learning model
Picture
Emotion prediction
Picture

Valence perception is the rapid, unconscious emotional reaction to images.

By predicting valence perception you can predict human behavior.
IMAGE PERCEPTION INVOLVES MORE THAN OBJECT RECOGNITION

A feature space to predict visual trends

Training a Convolutional Deep Neural Network (CDNN) on valence data generates a new type of feature space:
  • Images cluster based on style
  • Different objects with similar style are located nearby in the feature space
  • Tracking the location of style clusters over time allows you to model emergent trends
  • ​Custom models can be created that capture an individual's or a brand's visual identity
Picture

Ready to try?

Developers

About

Team
Medium
​GitHub
License
Patents



Contact

Google Group
Twitter
​Instagram
Email
© NEON OPEN 2017
  • Use cases
  • How it works
  • Developers