Benchmarking Ultra-High-Definition Image Super-resolution

Abstract

Deep learning based methods have achieved significant success in the task of single image reflection removal (SIRR). However, the majority of these methods are focused on High-Definition/Standard-Definition (HD/SD) images, while ignoring higher resolution images such as Ultra-High-Definition (UHD) images. With the increasing prevalence of UHD images captured by modern devices, in this paper, we aim to address the problem of UHD SIRR. Specifically, we first synthesize two large-scale UHD datasets, UHDRR4K and UHDRR8K. The UHDRR4K dataset consists of 2, 999 and 168 quadruplets of images for training and testing respectively, and the UHDRR8K dataset contains 1, 014 and 105 quadruplets. To the best of our knowledge, these two datasets are the first largest-scale UHD datasets for SIRR. Then, we conduct a comprehensive evaluation of six state-of-the-art SIRR methods using the proposed datasets. Based on the results, we provide detailed discussions regarding the strengths and limitations of these methods when applied to UHD images. Finally, we present a transformerbased architecture named RRFormer for reflection removal. RRFormer comprises three modules, namely the Prepossessing Embedding Module, Self-attention Feature Extraction Module, and Multi-scale Spatial Feature Extraction Module. These modules extract hypercolumn features, global and partial attention features, and multi-scale spatial features, respectively. To ensure effective training, we utilize three terms in our loss function: pixel loss, feature loss, and adversarial loss. We demonstrate through experimental results that RRFormer achieves stateof-the-art performance on both the non-UHD dataset and our proposed UHDRR datasets. The code and datasets are publicly available at https://github.com/Liar-zzy/Benchmarking-Ultra-High-Definition-Single-Image-Reflection-Removal.

Slice-Aided Defect Detection in Ultra High-Resolution Wind Turbine Blade Images

“…High-resolution typically refers to images with dimensions of 1920 × 1080 pixels, whereas ultra high-resolution starts at 3840 × 2160 pixels. The images we use here have a resolution of 5280 × 2890 pixels, firmly placing them in the ultra high-resolution category [36]. Because DTU does not provide surface defect bounding boxes, we initiated an annotation process after conducting a study of the different types of wind turbine blade surface defects noticeable in the database.…”

Study of Deep Learning-based models for Single Image Super-Resolution

“…Several works have performed benchmarks for SR, but they address this problem from different angles, which remains useless for users to guide their choice of model for SR. For example, Chen et al focused on real-world single image superresolution (RSISR) to address the problem of SR degradation on synthetic data [19]. Other works are not recent and address methods that are outdated today as the case for [23,33]. Zhang et al [34] worked on Ultra High Definition (UHD), introducing two datasets UHDSR4K and UHDSR8K which remain limited to this type of image.…”

Comparing the Robustness of Modern No-Reference Image- and Video-Quality Metrics to Adversarial Attacks

“…For example, despite LPIPS already being shown to be vulnerable to adversarial attacks, it is still used as the main metric in some benchmarks, e.g. to compare super-resolution methods (Zhang et al 2021). In some competitions that publish the results of subjective comparisons and objective quality scores, we can see the vast difference in these leaderboards.…”