Half a century ago, the Apollo 11 mission realized humanity’s dream of landing on the moon. To celebrate the 50th anniversary of the Apollo moon landing, we are launching TensorFlow’s latest space game, “Mission to TensorFlow World”.
This game is based on the TensorFlow product ecosystem: TensorFlow Core, TensorFlow.js, TensorFlow Lite, and TensorFlow Extended (TFX), controlling a spaceship navigating through vast space. At each space station, you must use different platforms and versions of TensorFlow to operate the spaceship, and these actions will also affect the main visual effects.
Pose Shield (TensorFlow.js)
During the “Mission to TensorFlow World”, the spaceship needs to pass through a dangerous asteroid belt. Using PoseNet and TensorFlow.js, you can enable or disable the shield around the spaceship by raising your arms to block meteors. You need to be agile and precise enough to protect the ship from damage; visit g.co/PoseShield to play the game.
Emojicules (TensorFlow Lite)
No space travel can avoid mechanical failures, and unfortunately, the spaceship encountered cabin depressurization during this flight. You need to solve the problem at hand using emotion classification and TensorFlow Lite.
You must make expressions corresponding to the emojis to capture various molecules (such as nitrogen, oxygen, carbon dioxide, and argon) in the breathable air they represent, in order to restore cabin pressure as quickly as possible. Visit the codebase (https://github.com/google/emojicules-tflite) for more information.
Interstellar Mapping (TensorFlow Core, TFX)
As the “Mission to TensorFlow World” progresses, the spaceship enters a new galaxy, and mapping the route is extremely challenging. In this challenge, you need to use XLNet and TensorFlow Core to select the statements represented by the planets; you can only pass through when the statements match. TFX is an important validation tool for all TensorFlow implementations.
Interstellar mapping first plays an animation to introduce the purpose and usage of the components. Once in the game, the TFX calibration animation will showcase the new planetary clusters, using TFX’s ability to identify and discard erroneous data. See how many galaxies you can pass through before time runs out! Visit the codebase (https://github.com/google/startographer-tfcore) for more information.
If you can successfully reach TensorFlow World, we will see you again soon! If you cannot make it, then we’ll see you live~
