ITM-HDR-VQA DATASET

The ITM-HDR-VQA dataset is a video quality assessment dataset for inversely tone-mapped videos. It contain 200 HDR10 videos with their MOS.
We capture videos of 20 typical HDR scenes including daylight scenes containing both sunlit areas and deep shadows and night scenes lit by artificial lights. The contents of these scenes can be roughly divided into two categories, man-made architecture and natural scenery.
A SONY A7S3 camera capable of recording a luminance range of up to 15 stops was used. The camera configurations are as follows: conversion curve = S-Log3, color management space = S-Gamut3, resolution = 3840×2160, and frame rate = 50 fps. After obtaining SDR videos, we map them to ITM-HDR-Videos through ten representative ITM algorithms, including traditional and deep learning-based methods.