BVI-RLV: A Fully Registered Dataset and Benchmarks for Low-Light Video Enhancement
This provides a valuable resource for researchers in computer vision working on low-light video enhancement, though it is incremental as it focuses on dataset creation rather than a new method.
The paper tackles the problem of low-light video enhancement by introducing a fully registered dataset with 40 scenes under two low-light conditions, and benchmarks show models trained on it outperform existing datasets.
Low-light videos often exhibit spatiotemporal incoherent noise, compromising visibility and performance in computer vision applications. One significant challenge in enhancing such content using deep learning is the scarcity of training data. This paper introduces a novel low-light video dataset, consisting of 40 scenes with various motion scenarios under two distinct low-lighting conditions, incorporating genuine noise and temporal artifacts. We provide fully registered ground truth data captured in normal light using a programmable motorized dolly and refine it via an image-based approach for pixel-wise frame alignment across different light levels. We provide benchmarks based on four different technologies: convolutional neural networks, transformers, diffusion models, and state space models (mamba). Our experimental results demonstrate the significance of fully registered video pairs for low-light video enhancement (LLVE) and the comprehensive evaluation shows that the models trained with our dataset outperform those trained with the existing datasets. Our dataset and links to benchmarks are publicly available at https://doi.org/10.21227/mzny-8c77.