Meta Platforms, Inc.
Image compression using content categories

Abstract:

In one embodiment, a method includes analyzing content of an image, identifying a category of the image based on the analyzed content of the image, identifying, based on the category of the image, a set of one or more compression parameters for a compression algorithm, and compressing the image using the compression algorithm, wherein the compression parameters are provided as input to the compression algorithm. The category of the image may include a category of an object depicted in the image, and identifying the category of the image may include detecting one or more objects in the image, selecting a primary subject of the image from the objects in the image, and identifying a category of the primary subject, wherein the category of the image comprises the category of the primary subject. The primary subject may be the largest of the objects in the image.