![]() ![]() The pipeline lived on Google Compute Engine and, at first, consisted of a set of Node.js scripts that could query BigQuery and output data to Google Cloud Storage. In order to supply David with the data to train his model, we created a pipeline for processing the data from BigQuery into a simplified form that would be easier for him to consume. One of the demos from the SketchRNN playground David ported the SketchRNN model to Javascript and built several inspiring demos that you can play with live in the browser. You can try it out yourself as it’s been open sourced via TensorFlow Magenta. He developed SketchRNN, a neural network that has learned how to draw words and can do other interesting things like interpolate between drawings. One of the first research projects to come out of the dataset was done by David Ha of the Google Brain team. Once we started doing some basic analysis of the dataset, it became clear to us that it could offer some fascinating opportunities for more advanced research. The full dataset of more than one billion vector drawings and metadata weighs in at over 3TB of data, which is no problem for BigQuery. Every five minutes, a cron job running on App Engine copies data from Datastore into BigQuery. In order to ask questions about the data - from simple things like how many drawings exist for each word, to more complex inquiries into similarities between the way people represent certain concepts - we needed to move the data into a more powerful tool. When it came time to start analyzing the data though, we looked to another tool, BigQuery. Due to its flexible and scalable nature, the team never had to worry about how many drawings it was saving having an impact on the game's performance. The data in JSON format of the 1 billionth drawingĭatastore ended up being an excellent choice. The data needed to reconstruct the drawing was saved alongside the metadata in the same Datastore object, in the form of arrays of x,y and time coordinates. The team was very careful to respect user privacy so it only saved anonymous metadata, including timestamp, country code, whether or not the drawing was recognized, and which word the drawing corresponded to. Although our game uses an internal API, the Handwriting team has released a TensorFlow tutorial which you can follow to recreate the model.įrom the beginning of the project we thought a collection of anonymous drawings could lead to an interesting research dataset, so we saved each drawing into a Datastore collection once the user was finished drawing it. Hand drawn doodles have a lot in common with handwritten words, so the Handwriting team was able to customize one of their models to recognize the game's drawings. ![]() This is made possible by a machine learning model built by the Handwriting team, part of the Machine Perception team based in the Mountain View and Zurich offices. Most people tell us their favorite moment in the game is when the AI guesses your drawing correctly. Even with over 30,000 people playing at once, App Engine automatically scaled up instances to meet the demand. Furthermore, the team found they didn't have to do any extra infrastructure work after they launched. App Engine’s flexibility and ease-of-use allowed them to focus on making the game fun. The original Quick, Draw! game was developed and deployed on App Engine by Jonas Jongejan on the Google Creative Lab team. Read on to learn how the game was built, and how you can get your hands on what’s arguably the world’s cutest research dataset. But it was all made possible through Google Cloud Platform services like App Engine, Cloud Datastore, and BigQuery. Quick, Draw! was brought to life through a collaboration between artists, designers, developers and research scientists from different teams across Google. We’ve now released 50 million of those drawings, and we’re thrilled that it’s stirred interest in the ML research community as well as the general public. Over a billion doodles have been drawn by people playing the game and subsequently collected into this anonymized dataset. ![]() Along the way, we’ve had the opportunity to help teach a neural network to recognize drawings, plus generate the world’s largest doodling data set, which is shared publicly to help further machine learning research. The game asks users to draw a doodle, then the game’s AI tries to guess what it is. Quick, Draw! is an AI experiment that has delighted millions of people across the world. ![]()
0 Comments
Leave a Reply. |