Submissions by ianmoh tagged neural-networks

Mosaic is a puzzle-based time attack where players watching blurred images, guess the name, subject, and answer other questions about it, done under a pressure of time.

In starting a new game, a player encounters a heavily distorted image. A countdown begins from a predetermined amount, and the picture proceeds to un-blur, growing sharper in fidelity and contrast, until the original image is visible again. Players have until the time runs out to answer questions concerning the image's specific details. These questions typically reflect:

  • People or objects in view
  • Presence of landscape objects such as land, sea, sky
  • Emotion
  • Famous landmarks

Each level loads a different image, which progressively becomes more abstract and specific, covering esoteric or advanced questions. For instance, fine art paintings with advanced questions normally relegated to armchair theologians with specialised knowledge.

You are awarded points based on how many questions you answered. The point scale multiplies and rewards you the more questions are answer, effectively being a measure of how well and quickly you can discern the blurred images.

The inspiration behind this game is from the use of Captcha. Captcha verifies against bots (non-human users) by getting them identify augmented words and letters in images. As well as this, this is used to aid a deep learning algorithm is optical recognition of printed text.

This game applies the idea of gamification, to crowd source the info for machine of software to understand images in a more intuitive sense – for instance emotional nuances, wider context (two people getting married at a wedding), further to objective and literal details (tree, man, woman). Ultimately, this presence a novel way to develop Artificial General Intelligence (A.G.I.).

Platform: Desktop platform using Facebook as a medium
Target Audience: Young adults 18 years and older

Reference Image:
http://thedsgnblog.com