1. Introduction
Social media filters (e.g. Instagram filters) transform an image into a different version by applying several transformations, and this modified version may have color-level or pixel-level corruptions and perturbations. These filters modify the original image by adjusting the contrast and brightness, or changing hue and saturation, or introducing different levels of blur and noise, or applying color curves or vignetting. Though these filters convert images to a more aesthetically pleasing appearance, they also make the content in those images more complicated to understand by learning-based algorithms. Therefore, removing the filters from social media images is a crucial preprocessing step completed for the visual analysis of social media contents.