Benchmarking Low-Light Image Enhancement and Beyond

Abstract

In this paper, we present a systematic review and evaluation of existing single-image low-light enhancement algorithms. Besides the commonly used low-level vision oriented evaluations, we additionally consider measuring machine vision performance in the low-light condition via face detection task to explore the potential of joint optimization of high-level and low-level vision enhancement. To this end, we first propose a large-scale low-light image dataset serving both low/high-level vision with diversified scenes and contents as well as complex degradation in real scenarios, called Vision Enhancement in the LOw-Light condition (VE-LOL). Beyond paired low/normal-light images without annotations, we additionally include the analysis resource related to human, i.e. face images in the low-light condition with annotated face bounding boxes. Then, efforts are made on benchmarking from the perspective of both human and machine visions. A rich variety of criteria is used for the low-level vision evaluation, including full-reference, no-reference, and semantic similarity metrics. We also measure the effects of the low-light enhancement on face detection in the low-light condition. State-of-the-art face detection methods are used in the evaluation. Furthermore, with the rich material of VE-LOL, we explore the novel problem of joint low-light enhancement and face detection. We develop an enhanced face detector to apply low-light enhancement and face detection jointly. The features extracted by the enhancement module are fed to the successive layer with the same resolution of the detection module. Thus, these features are intertwined together to unitedly learn useful information across two phases, i.e. enhancement and detection. Experiments on VE-LOL provide a comparison of state-of-the-art low-light enhancement algorithms, point out their limitations, and suggest promising future directions. Our dataset has supported the Track "Face Detection in Low Light Conditions" of CVPR UG2+ Challenge (2019-2020) (http://cvpr2020.ug2challenge.org/).

IceNet for Interactive Contrast Enhancement

“…The objectives of enhancement are closely related but different between CE [15], [17], color enhancement [18], dehazing [19]- [21], and detail enhancement [22]. This section briefly reviews CE techniques.…”

“…These parametric curves produce promising results, but it is challenging to find reliable parameters, which are effective for diverse images. Recently, with the success of convolutional neural networks (CNNs) in the field of low-level vision [11]- [14], CNN-based CE methods also have been proposed, yielding outstanding performance [15].…”

Low-Light Image Enhancement with Normalizing Flow

“…It is a large-scale dataset including 2500 paired images with more diversified scenes and contents, thus is valuable for the cross-dataset evaluation. 1) Cross-dataset evaluation: We first evaluate the generality of our method in a cross-dataset manner, i.e., we train our method on the LOL dataset (Wei et al 2018) and test the model on the testing set of VE-LOL dataset (Liu et al 2021a).…”

Intrinsic Image Transfer for Illumination Manipulation

“…More recently, deep learning methods on unpaired datasets for image enhancement and tone mapping [10], [28], inspired by unsupervised image-to-image translation [59], bring new hope for easy and flexible image enhancement on real-world low-light images. The reader is referred to a systematic review and evaluation of existing deep learning-based low-light enhancement algorithms [37].…”