Brighten

This synthetic image calibration experiment is inspired by DeepFL-IQA: Weak Supervision for Deep IQA Feature Learning.

We manually set the parameter values that control the distortion amount such that the perceptual visual quality of the distorted images varies linearly with the distortions parameter, from an expected rating of 1 (bad) to 5 (excellent). The distortion parameter values were chosen based on a small set of images and then applied to all images in both datasets.