Tidy Up!
Sort clutter into matching bins, then wipe every surface clean before the two-minute timer runs out.
Independent game guide
Draw six prompted objects while a neural network tries to recognize each doodle before time runs out.
Source checked by GameWren.The game opens on the creator’s own page; GameWren does not rehost it.
Quick, Draw! gives you an object and twenty seconds to sketch it. A neural network watches each stroke and calls out guesses while you draw. If it recognizes the prompt, the round ends early; if not, you still move on to the next object in the set.
Simple silhouettes usually communicate faster than careful detail. Draw the feature that separates the object from similar things first: a handle, a wheel, a beak, or another strong clue. The spoken guesses show what the model thinks it sees and can tell you when the drawing is heading in the wrong direction.
Google describes the project as both a game and an example of machine learning. The model does not understand a drawing exactly as a person does; it compares the strokes with patterns learned from many other submitted doodles.
Those drawings were collected into a public dataset for research and creative projects. That makes every short round a compact demonstration of recognition, training data, and error: the network can identify remarkably rough sketches, but an unusual visual shorthand can still confuse it.
Can a neural network learn to recognize doodling?
Video game AI — A neural network guesses each drawing stroke by stroke.
Sort clutter into matching bins, then wipe every surface clean before the two-minute timer runs out.
A word-association game where machine-learned language understanding matches your clue to the words in play.
Place ramps, gates, fans, burners, and other tools, then steer flowing materials into a vessel without mixing in hazards.
Read a target truth table, place the missing logic gates, flip the inputs, and make every row turn green to light the lamp.