-
Notifications
You must be signed in to change notification settings - Fork 3
Home
KT edited this page Apr 15, 2018
·
1 revision
PerkinsHacks18 wiki!
Challenge:
Designing a solution:
-
Design Thinking Approach
-
learning from the users, thier expectation, user centric, tailored the solution to incorporate what they needed, e.g. frequency beeps instead of text to speech instructions
-
conversational bot can switch between beep and voice commands
-
Brainstorming
How it works? - The underlying Idea
- conversational AI
- speech to text to speech system
- computer vision
- mapping a touchscreen into grid co-ordinates
- differentiating static display buttons and digital display [which shows variable output]
- event recognition [actual press on a touchscree should change the display - an event is fired] by detecting changes in the digital panel
How to extend the prototype into a mobile app ?
Overcoming the technical challenges:
- platform independent vision algorithms - so that they can be used from any camera
- algorithms for mapping a screen display
- Hand profiling and hand detection customized to the user, deep learning module for calliberation
- illumination independence, reducing signal to noise ratio, handling reflections
- smooth and more intutive beep generation
- automatic display calliberation and grid map generation using vision and OCR
- allignment, orientation, display recognition on an applicance (like Microsoft's seeing AI)
- proximity and distance analysis for measuring closeness to the digital screen