Inspiration
Usually stores lack a way to track customer demographics. With our software built using opencv and IBM Watson's visual recognition api, customer profiles can be kept to improve add insights.
What it does
Uses opencv to track faces and eyes from a video source.
How we built it
In python, a video source is read and analyzed for faces and eyes in real-time. Frame by frame the video is transferred into gray scale, structures are highlighted and then compared with pre-determined shapes. If a frame is recognized to contain a face, it is then scanned for eyes. Faces and eye highlights are then presented in an overlay.. .
Challenges we ran into
- lack of sleep
- raspberry pi issues resulting in a lockout
- library and module compatibility
- installation issues
- documentation issues
- webcam incompatibility
- slow / crashing internet
Accomplishments that we're proud of
The software is working and able to track faces and eyes with better accuracy then expected.
What we learned
- cool new libraries
- raspbian
- matplotlib
- opencv
- tkinter
- watson_developer_cloud
- better planning and preparation is required
- software is hard
- hardware is even harder
What's next for inSIGHT
Future implementations with IBM Watson's facial recognition api would be able to build fully detailed customer profiles; what has been previously ordered; what is likely to be ordered; how often does a customer come back; probability of a customer returning.
Log in or sign up for Devpost to join the conversation.