Christos G. Bampis

Li-Heng Chen, Christos G. Bampis, Zhi Li et al.

We describe a search-free resizing framework that can further improve the rate-distortion tradeoff of recent learned image compression models. Our approach is simple: compose a pair of differentiable downsampling/upsampling layers that sandwich a neural compression model. To determine resize factors for different inputs, we utilize another neural network jointly trained with the compression model, with the end goal of minimizing the rate-distortion objective. Our results suggest that "compression friendly" downsampled representations can be quickly determined during encoding by using an auxiliary network and differentiable image warping. By conducting extensive experimental tests on existing deep image compression models, we show results that our new resizing parameter estimation framework can provide Bjøntegaard-Delta rate (BD-rate) improvement of about 10% against leading perceptual quality engines. We also carried out a subjective quality study, the results of which show that our new approach yields favorable compressed images. To facilitate reproducible research in this direction, the implementation used in this paper is being made freely available online at: https://github.com/treammm/ResizeCompression.

Zhi Li, Christos G. Bampis, Lukáš Krasula et al.

In a subjective experiment to evaluate the perceptual audiovisual quality of multimedia and television services, raw opinion scores collected from test subjects are often noisy and unreliable. To produce the final mean opinion scores (MOS), recommendations such as ITU-R BT.500, ITU-T P.910 and ITU-T P.913 standardize post-test screening procedures to clean up the raw opinion scores, using techniques such as subject outlier rejection and bias removal. In this paper, we analyze the prior standardized techniques to demonstrate their weaknesses. As an alternative, we propose a simple model to account for two of the most dominant behaviors of subject inaccuracy: bias and inconsistency. We further show that this model can also effectively deal with inattentive subjects that give random scores. We propose to use maximum likelihood estimation to jointly solve the model parameters, and present two numeric solvers: the first based on the Newton-Raphson method, and the second based on an alternating projection (AP). We show that the AP solver generalizes the ITU-T P.913 post-test screening procedure by weighing a subject's contribution to the true quality score by her consistency (thus, the quality scores estimated can be interpreted as bias-subtracted consistency-weighted MOS). We compare the proposed methods with the standardized techniques using real datasets and synthetic simulations, and demonstrate that the proposed methods are the most valuable when the test conditions are challenging (for example, crowdsourcing and cross-lab studies), offering advantages such as better model-data fit, tighter confidence intervals, better robustness against subject outliers, the absence of hard coded parameters and thresholds, and auxiliary information on test subjects. The code for this work is open-sourced at https://github.com/Netflix/sureal.

Christos G. Bampis, Zhi Li, Alan C. Bovik

Perceptual video quality assessment models are either frame-based or video-based, i.e., they apply spatiotemporal filtering or motion estimation to capture temporal video distortions. Despite their good performance on video quality databases, video-based approaches are time-consuming and harder to efficiently deploy. To balance between high performance and computational efficiency, Netflix developed the Video Multi-method Assessment Fusion (VMAF) framework, which integrates multiple quality-aware features to predict video quality. Nevertheless, this fusion framework does not fully exploit temporal video quality measurements which are relevant to temporal video distortions. To this end, we propose two improvements to the VMAF framework: SpatioTemporal VMAF and Ensemble VMAF. Both algorithms exploit efficient temporal video features which are fed into a single or multiple regression models. To train our models, we designed a large subjective database and evaluated the proposed models against state-of-the-art approaches. The compared algorithms will be made available as part of the open source package in https://github.com/Netflix/vmaf.

Dounia Hammou, Yancheng Cai, Pavan Madhusudanarao et al.

Image and video quality metrics, such as SSIM, LPIPS, and VMAF, are aimed to predict the perceived quality of the evaluated content and are often claimed to be "perceptual". Yet, few metrics directly model human visual perception, and most rely on hand-crafted formulas or training datasets to achieve alignment with perceptual data. In this paper, we propose a set of tests for full-reference quality metrics that examine their ability to model several aspects of low-level human vision: contrast sensitivity, contrast masking, and contrast matching. The tests are meant to provide additional scrutiny for newly proposed metrics. We use our tests to analyze 33 existing image and video quality metrics and find their strengths and weaknesses, such as the ability of LPIPS and MS-SSIM to predict contrast masking and poor performance of VMAF in this task. We further find that the popular SSIM metric overemphasizes differences in high spatial frequencies, but its multi-scale counterpart, MS-SSIM, addresses this shortcoming. Such findings cannot be easily made using existing evaluation protocols.

Lukáš Krasula, Zhi Li, Christos G. Bampis et al.

Staircase-like contours introduced to a video by quantization in flat areas, commonly known as banding, have been a long-standing problem in both video processing and quality assessment communities. The fact that even a relatively small change of the original pixel values can result in a strong impact on perceived quality makes banding especially difficult to be detected by objective quality metrics. In this paper, we study how banding annoyance compares to more commonly studied scaling and compression artifacts with respect to the overall perceptual quality. We further propose a simple combination of VMAF and the recently developed banding index, CAMBI, into a banding-aware video quality metric showing improved correlation with overall perceived quality.

Li-Heng Chen, Christos G. Bampis, Zhi Li et al.

The layers of convolutional neural networks (CNNs) can be used to alter the resolution of their inputs, but the scaling factors are limited to integer values. However, in many image and video processing applications, the ability to resize by a fractional factor would be advantageous. One example is conversion between resolutions standardized for video compression, such as from 1080p to 720p. To solve this problem, we propose an alternative building block, formulated as a conventional convolutional layer followed by a differentiable resizer. More concretely, the convolutional layer preserves the resolution of the input, while the resizing operation is fully handled by the resizer. In this way, any CNN architecture can be adapted for non-integer resizing. As an application, we replace the resizing convolutional layer of a modern deep downsampling model by the proposed building block, and apply it to an adaptive bitrate video streaming scenario. Our experimental results show that an improvement in coding efficiency over the conventional Lanczos algorithm is attained, in terms of PSNR, SSIM, and VMAF on test videos.

Li-Heng Chen, Christos G. Bampis, Zhi Li et al.

Mean squared error (MSE) and $\ell_p$ norms have largely dominated the measurement of loss in neural networks due to their simplicity and analytical properties. However, when used to assess visual information loss, these simple norms are not highly consistent with human perception. Here, we propose a different proxy approach to optimize image analysis networks against quantitative perceptual models. Specifically, we construct a proxy network, which mimics the perceptual model while serving as a loss layer of the network.We experimentally demonstrate how this optimization framework can be applied to train an end-to-end optimized image compression network. By building on top of a modern deep image compression models, we are able to demonstrate an averaged bitrate reduction of $28.7\%$ over MSE optimization, given a specified perceptual quality (VMAF) level.

Li-Heng Chen, Christos G. Bampis, Zhi Li et al.

The use of $\ell_p$ $(p=1,2)$ norms has largely dominated the measurement of loss in neural networks due to their simplicity and analytical properties. However, when used to assess the loss of visual information, these simple norms are not very consistent with human perception. Here, we describe a different "proximal" approach to optimize image analysis networks against quantitative perceptual models. Specifically, we construct a proxy network, broadly termed ProxIQA, which mimics the perceptual model while serving as a loss layer of the network. We experimentally demonstrate how this optimization framework can be applied to train an end-to-end optimized image compression network. By building on top of an existing deep image compression model, we are able to demonstrate a bitrate reduction of as much as $31\%$ over MSE optimization, given a specified perceptual quality (VMAF) level.

Sungsoo Kim, Jin Soo Park, Christos G. Bampis et al.

We propose a video compression framework using conditional Generative Adversarial Networks (GANs). We rely on two encoders: one that deploys a standard video codec and another which generates low-level maps via a pipeline of down-sampling, a newly devised soft edge detector, and a novel lossless compression scheme. For decoding, we use a standard video decoder as well as a neural network based one, which is trained using a conditional GAN. Recent "deep" approaches to video compression require multiple videos to pre-train generative networks to conduct interpolation. In contrast to this prior work, our scheme trains a generative decoder on pairs of a very limited number of key frames taken from a single video and corresponding low-level maps. The trained decoder produces reconstructed frames relying on a guidance of low-level maps, without any interpolation. Experiments on a diverse set of 131 videos demonstrate that our proposed GAN-based compression engine achieves much higher quality reconstructions at very low bitrates than prevailing standard codecs such as H.264 or HEVC.

Christos G. Bampis, Alan C. Bovik

HTTP-based video streaming technologies allow for flexible rate selection strategies that account for time-varying network conditions. Such rate changes may adversely affect the user's Quality of Experience; hence online prediction of the time varying subjective quality can lead to perceptually optimised bitrate allocation policies. Recent studies have proposed to use dynamic network approaches for continuous-time prediction; yet they do not consider multiple video quality models as inputs nor consider forecasting ensembles. Here we address the problem of predicting continuous-time subjective quality using multiple inputs fed to a non-linear autoregressive network. By considering multiple network configurations and by applying simple averaging forecasting techniques, we are able to considerably improve prediction performance and decrease forecasting errors.

Christos G. Bampis, Alan C. Bovik

Mobile streaming video data accounts for a large and increasing percentage of wireless network traffic. The available bandwidths of modern wireless networks are often unstable, leading to difficulties in delivering smooth, high-quality video. Streaming service providers such as Netflix and YouTube attempt to adapt their systems to adjust in response to these bandwidth limitations by changing the video bitrate or, failing that, allowing playback interruptions (rebuffering). Being able to predict end user' quality of experience (QoE) resulting from these adjustments could lead to perceptually-driven network resource allocation strategies that would deliver streaming content of higher quality to clients, while being cost effective for providers. Existing objective QoE models only consider the effects on user QoE of video quality changes or playback interruptions. For streaming applications, adaptive network strategies may involve a combination of dynamic bitrate allocation along with playback interruptions when the available bandwidth reaches a very low value. Towards effectively predicting user QoE, we propose Video Assessment of TemporaL Artifacts and Stalls (Video ATLAS): a machine learning framework where we combine a number of QoE-related features, including objective quality features, rebuffering-aware features and memory-driven features to make QoE predictions. We evaluated our learning-based QoE prediction model on the recently designed LIVE-Netflix Video QoE Database which consists of practical playout patterns, where the videos are afflicted by both quality changes and rebuffering events, and found that it provides improved performance over state-of-the-art video quality metrics while generalizing well on different datasets. The proposed algorithm is made publicly available at http://live.ece.utexas.edu/research/Quality/VideoATLAS release_v2.rar.

Zhi Li, Christos G. Bampis

Simple quality metrics such as PSNR are known to not correlate well with subjective quality when tested across a wide spectrum of video content or quality regime. Recently, efforts have been made in designing objective quality metrics trained on subjective data (e.g. VMAF), demonstrating better correlation with video quality perceived by human. Clearly, the accuracy of such a metric heavily depends on the quality of the subjective data that it is trained on. In this paper, we propose a new approach to recover subjective quality scores from noisy raw measurements, using maximum likelihood estimation, by jointly estimating the subjective quality of impaired videos, the bias and consistency of test subjects, and the ambiguity of video contents all together. We also derive closed-from expression for the confidence interval of each estimate. Compared to previous methods which partially exploit the subjective information, our approach is able to exploit the information in full, yielding tighter confidence interval and better handling of outliers without the need for z-scoring or subject rejection. It also handles missing data more gracefully. Finally, as side information, it provides interesting insights on the test subjects and video contents.