Learning Flexible Generalization in Video Quality Assessment by Bringing Device and Viewing Condition Distributions
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors studied how people watch videos on many different Android phones, considering things like lighting and screen quality that affect how good the video looks. They collected a big set of real user opinions from over 300 devices to better understand these effects. They then developed a way to adjust video quality scores based on the specific device and viewing situation. Their work helps improve video quality measurements by making them more accurate for real-world use, beyond just lab tests. They also shared their data and tools for others to use.
Video Quality AssessmentSubjective DatasetViewing ConditionsDisplay CharacteristicsAndroid DevicesQuality PredictionStreaming OptimizationPerceptual Quality ModelsMetadata
Authors
Nikolay Safonov, Dmitriy S. Vatolin
Abstract
Video quality assessment (VQA) plays a critical role in optimizing video delivery systems. While numerous objective metrics have been proposed to approximate human perception, the perceived quality strongly depends on viewing conditions and display characteristics. Factors such as ambient lighting, display brightness, and resolution significantly influence the visibility of distortions. In this work, we address the question of the multi-screen quality assessment on mobile devices, as this area still tends to be under-covered. We introduce a first large-scale subjective dataset collected across more than different 300 Android devices, accompanied by metadata on viewing conditions and display properties. We propose a strategy for aggregated score extraction and adaptation of VQA models to device-specific quality estimation. Our results demonstrate that incorporating device and context information enables more accurate and flexible quality prediction, offering new opportunities for fine-grained optimization in streaming services. Ultimately, this work advances the development of perceptual quality models that bridge the gap between laboratory evaluations and the diverse conditions of real-world media consumption. We made the dataset and the code available at https://videoprocessing.github.io/device-viewing-conditions.