Picture This

connected frame with historic images

A connected picture frame displaying historic images

farm9.staticflickr.com_8681_16672996925_ec53c13c61.jpg

Story

  • The Picture This “smart” frame shows police photographs of homeless people by Carl Durheim (1810-1890)
  • By looking at a picture, you trigger a face detection algorithm to analyse both, you and the homeless person
  • The algorithm detects gender, age and mood of the person on the portrait (not always right)
  • You, as a spectator, become part of the system / algorithm judging the homeless person
  • The person on the picture is at the mercy of the spectator, once again

farm9.staticflickr.com_8631_16485421338_40e86f1bce.jpg farm9.staticflickr.com_8608_16671627211_074399694d.jpg

How it works

  • Picture frame has a camera doing face detection for presence detection
  • Pictures have been pre-processed using a cloud service
  • Detection is still rather slow (should run faster on the Raspi 2)
  • Here's a little video https://www.flickr.com/photos/tamberg/16053255113/

farm9.staticflickr.com_8608_16672997165_b32d138ee9.jpg farm9.staticflickr.com_8577_16671979502_b6a4cc5bd0.jpg farm9.staticflickr.com_8611_16673072845_aea4ee6a02_z.jpg

Questions (not) answered

  • Who were those people? Why were they homeless? What was their crime?
  • How would we treat them? How will they / we be treated tomorrow? (by algorithms)

Data

Team

  • @ram00n
  • @tamberg
  • and you

Ideas / Iterations

  1. Download the pictures to the Raspi and display one of them (warmup)
  2. Slideshow and turning the images 90° to adapt to the screensize
  3. Play around with potentiometer and Arduino to bring an analog input onto the Raspi (which only has digital I/O)
  4. Connect everything and adapt the slideshow speed with the potentiometer
  5. Display the name (extracted from the filename) below the picture

next steps, more ideas:

  1. Use the Raspi Cam to detect a visitor in front of the frame and stop the slideshow
  2. Use the Raspi Cam to take a picture of the face of the visitor
  3. Detect faces in the camera picture
  4. Detect faces in the images [DONE, manually, using online service]
  5. …merge visitor and picture faces :-)

Material

Software

Links

Not used this time, but might be useful

Event finish

Start