wwwconversation
2019www.conversation shows a different way of communication between human and technology.
Can machines love? What if we could talk to computers with our emotions? Would it look like a more human-like conversation?
This web project detects your emotions and makes the computer respond to them in a cute way.
P.S. Works for Google Chrome.
Roles
This is a solo project.
- Pretrained model by Kevin Hsiao,
- Code integration & visuals by Yona Ngo.
Visuals
The computer's 'face' was drawn in pixel art style, based on the idea, that pixel art represents everything techy or retro. The animations are powered by Pixaki app and p5.play. There are five expressions in total: neutral, happy, surprise, angry and fear.
How It Works
www.conversation uses a pretrained model by Kevin Hsiao to detect certain emotions: neutral, happy, angry, sad, disgust, fear and surprise. If emotion is detected, a background-color and animation change accordingly.
Tools & Process
Pretrained model via TensorFlow.js, visuals via p5.js and Pixaki.
Pretrained model via TensorFlow.js, visuals via p5.js and Pixaki.
Thank Thank Thank
Special thanks to Kevin Hsiao for letting me use his pretrained model for this project.
Special thanks to Kevin Hsiao for letting me use his pretrained model for this project.