A back limb will get colored like a spike. Some of my sketches aren't rule-conforming. Using strategic rule breaking and Pix2Pix to get to 100% The rest are spikes or petals, except for the dragon's eye, which can be distinguished by its distance from the background. The next biggest is the body (plus the arms and legs) or the flower's center. In a "rule-conforming" cartoon dragon or flower sketch, the biggest white component is the background. For example, when coloring a binary black and white image, if you click on a white pixel, the white pixels that are be reached without crossing over black are colored the new color. Although 80% isn't enough, we can bootstrap from that partial-rule-based solution to 100% using strategic rule-breaking transformations, augmentations, and machine learning.Ĭonnected components are what is colored when you use Windows Paint (or a similar application). But it turns out we can define the flower or dragon parts in terms of connected components and get a geometric solution for coloring about 80% of our drawings. Make the dragon's body orange and the spikes yellow.Īt first, that doesn't seem helpful because our computer doesn't know what a center or a petal or a body or a spike is. I could tell a kid how I want my drawings colored: Make the flower's center orange and the petals yellow. There is a non-machine-learning approach to solving my problem. I drew a few dozen cartoon flowers and dragons and asked whether I could somehow turn this into a training set. How many was I willing to draw? Maybe 30. The tens of thousands (or hundreds of thousands) of examples often required by deep-learning models were out of the question.īased on Pix2Pix's examples, we would need at least 400 to 1,000 sketch/colored pairs. I needed the training data to be very limited because I didn't want to draw and color a lifetime supply of cartoon characters just to train the model. I had seen Pix2Pix, a machine learning image-to-image translation model described in a paper ("Image-to-Image Translation with Conditional Adversarial Networks," by Isola, et al.), that colorizes landscapes after training on AB pairs where A is the grayscale version of landscape B. A borrowed TensorFlow image-to-image translation model, Pix2Pix, to automate cartoon coloring with very limited training data.A rule-based strategy for extreme augmentation of small datasets.In this experiment (which we called DragonPaint), we confronted the problem of deep learning's enormous labeled-data requirements using: The specific problem our group set out to solve was: Can we train a model to automate applying a simple color scheme to a black and white character without hand-drawing hundreds or thousands of examples as training data? Paying to label every frame of a movie-length video adds up fast, even at a penny a frame. Other times the volume of data needed multiplied by the cost of human labeling by Amazon Turkers or summer interns is just too high. Sometimes the domain is one in which there just isn't much data (for example, when diagnosing a rare disease or determining whether a signature matches a few known exemplars). It's a big problem especially if you don't have the labeled data-and even in a world awash with big data, most of us don't.Īlthough a few companies have access to enormous quantities of certain kinds of labeled data, for most organizations and many applications, creating sufficient quantities of the right kind of labeled data is cost prohibitive or impossible. Welcome to the communityĪ big problem with supervised machine learning is the need for huge amounts of labeled data.
0 Comments
Leave a Reply. |