Image colorization using machine learning

Abstract

Implementations described herein relate to methods, systems, and computer-readable media to train and use a machine-learning model to colorize a grayscale image that depicts a person. In some implementations, a computer-implemented method includes receiving the grayscale image. The method further includes generating a colorized image based on the grayscale image as output of a trained convolutional neural network (CNN) by providing the grayscale image as input to the trained CNN. In some implementations, the trained CNN performs part segmentation to detect one or more parts of the person and colorizes the grayscale image.

Description

Background technique [0001] Since cameras first became available, users have captured photos to preserve memories. For example, in the early days of photography, photographs were captured in black and white due to limitations in camera technology, due to the high cost of capturing color photographs, and so on. Subsequently, as color photos became available and cheaper, users took and stored color photos. [0002] Modern imaging applications allow users to store, view and edit photos. Many users have scanned their libraries for copies of old black and white photos as well as recently captured color photos. Users may like old black and white photos if they are rendered in color. [0003] While image colorization techniques, including machine learning-based techniques, can be used to colorize black-and-white images, it is difficult to get good output when applying such techniques to old black-and-white photographs, especially those depicting one or more people color image. F...

AI Technical Summary

Problems solved by technology

[0003] While image colorization techniques, including machine learning-based techniques, can be used to colorize black-and-white images, it is difficult to get good output when applying such techniques to old black-and-white photographs, especially those depicting one or more people color image

For example, bugs such as skin areas not matching the skin tone color; one skin area, e.g., the face gets color, while another, e.g., hands doesn't; etc. make the coloring of the photo suboptimal

[0004] Additionally, current image colorization techniques do not take into account the fact that people depicted in older black and white photographs often wear a different style of clothing than those in newer color photographs

Current image colorization techniques cannot account for this discrepancy when adding color to a black-and-white photograph depicting one or more people

Images