Jump to content

Quick, Draw!

From Wikipedia, the free encyclopedia
This article is about the drawing game. For similar terms, see Quickdraw (disambiguation).
Quick, Draw!
PublisherGoogle LLC
DesignersJonas Jongejan, Henry Rowley, Takashi Kawashima, Jongmin Kim, Ruben Thomson, Nick Fox-Gieg[1]
PlatformBrowser
ReleaseNovember 14, 2016
GenreGuessing game

Quick, Draw! is an online guessing game developed by the Google Creative Lab and Data Arts Team and published by Google LLC that challenges players to draw a picture of an object or concept, then uses a neural network-based artificial intelligence (AI) to guess what the drawings represent.[2][3][4] The AI uses an open-source dataset of all previous drawings to "learn" and improve its ability to guess correctly in future matches.[3] The game is similar to Pictionary in the sense that the player has a limited amount of time to draw (20 seconds).[2] The concepts that it guesses can be simple, like "circle", or more complicated, like "kangaroo ".[4]

Gameplay

[edit]
Bold text top center reading "Well drawn!" Body text below reading "Our neural net figured out 5 of your doodles. But it saw something else in the other 1. Select one to see what it saw, and visit the data to see 50 million drawings made by other real people on the internet." Below the text is the six drawings from the game, a couch, a duck, a hockey stick, trousers, a megaphone and a book. The game guessed five correctly, only missed the duck.
The results of a completed Quick, Draw! game on Sunday April 19, 2026.

In a match of Quick, Draw!, there are a total of six rounds. During each round, the player is given 20 seconds to draw a random prompt selected from the game's database whilst the AI attempts to guess the drawing. A round ends either when the AI successfully guesses the drawing or time runs out.

At the end of a Quick, Draw! match, the player is given their drawings and results for each round. When clicking on a drawing, the player can view the AI's comparisons of their drawing with other player-given drawings. From there, the player can share their drawings to Facebook or X, or replay the game, or quit.

Dataset

[edit]

The dataset of 50 million player drawings is openly available on the Quick, Draw! website. The site provides a step-by-step process of how the images are used by the AI and stored on the Quick, Draw! servers. The code collects the prompt given to the player, the drawing, whether the AI correctly guessed the drawing correct, and the country the player lives in[5]. This two-letter country code is the only piece of personal data the game saves from its players. The dataset stores the drawings of players through the Ramer–Douglas–Peucker algorithm, which simplifies the curves of the drawing so that the dataset stores less data points. The site also links the legal license that allows them to share, store and adapt data from the game.

Artificial Intelligence Training

[edit]

The Creative Team at Google used self-supervised representation learning to train the AI for Quick, Draw! For this, the team fed their machine a set of pre-labeled drawings, giving the AI a basis on which it can guess that drawing in the future.[6][7] In Quick, Draw!, the self-supervision comes from the dataset of previous drawings.[6][8] This means that all the developers need to do to continue training the AI is keep the site active and let people play the game.[9]

Data applications

[edit]
  • The Quick, Draw! dataset was used to train part of the app Spoken, which features a canvas that can recognize drawings and convert them to synthesized speech as a communication aid.
  • The technology used to recognize the doodles in Quick, Draw! is also used to recognize characters and languages by Google Translate. [2]

See also

[edit]

References

[edit]
  1. ^ "quick-draw". Retrieved 28 November 2016.
  2. ^ a b c Burgess, Matt (16 November 2016). "You can now play a Pictionary-style game called Quick Draw against Google's AI". Wired UK. Wired.co.uk. Retrieved 4 November 2024.
  3. ^ a b Lu, Wendy (23 November 2016). "How Does Google "Quick, Draw!" Work? This Game Makes Learning About Artificial Intelligence Fun". Bustle.com. Retrieved 4 November 2024.
  4. ^ a b Capewell, Jillian (21 November 2016). "Let A Computer Guess What You're Drawing In This High-Tech Pictionary Game". HuffingtonPost.com: Huffington Post. Retrieved 21 November 2016.
  5. ^ Johnson, Ian (2018-09-28). "Machine Learning for Visualization". Medium. Retrieved 2026-04-29.
  6. ^ a b Xu, Peng; Song, Zeyu; Yin, Qiyue; Song, Yi-Zhe; Wang, Liang (February 3, 2020). "Deep Self-Supervised Representation Learning for Free-Hand Sketch". Institute of Electrical and Electronics Engineers – via arXiv.
  7. ^ Lanham, Micheal (2020). Hands-On Reinforcement Learning for Games. Packt Publishing Ltd. (published January 2020). ISBN 978-1-83921-493-6.
  8. ^ "What is Self-Supervised Learning? | Stanford HAI". hai.stanford.edu. Retrieved 2026-04-29.
  9. ^ Xin, Zhang; Liangxiu, Han (November 3, 2023). "A Generic Self-Supervised Learning (SSL) Framework for Representation Learning from Spectral Spatial Features of Unlabeled Remote Sensing Imagery". Remote Sensing. 15 (21). doi:10.3390/r (inactive 30 April 2026). ISSN 2072-4292. Archived from the original on 2024-07-12 – via MDPI.{{cite journal}}: CS1 maint: DOI inactive as of April 2026 (link)
[edit]
Retrieved from "https://en.wikipedia.org/w/index.php?title=Quick,_Draw!&oldid=1355743727"