top of page
eXpressMe!
Road Map
01
Further refine action detection model
We would like to explore and establish a more consistent and stable action detection model - in conjunction with a medium degree of sensitivity to actioned captured in short frames within a video.
02
Optimize / reinforce sentence generation algo
To further enrich our sentence generation algorithm, in combination with a contextual extension to fully create an immersive dialogue. Would like to leverage reinforcement learning which is based on rewarding desired behaviors and/or punishing undesired ones.
03
Better extract emotion from non-verbal human sound
Despite we can successfully capture and interpret non-verbal sound like clapping, yawning and other human sounds, incorporating interpreted emotion into day-to-day scenario posed another challenge to our application.
bottom of page