Video Quality Metric

Video quality metrics aim to objectively assess the perceptual quality of videos, crucial for benchmarking video processing algorithms and optimizing user experience. Current research heavily focuses on developing robust no-reference metrics, particularly those based on deep learning, while simultaneously addressing their vulnerability to adversarial attacks that artificially inflate scores. This work is significant because reliable metrics are essential for fair algorithm comparison and for guiding the development of improved video compression and processing techniques across diverse applications, including omnidirectional video and user-generated content. The development of specialized metrics tailored to specific video processing tasks, such as frame interpolation, is also a growing area of interest.