Low-Light Enhancement Effect on Classification and Detection: An Empirical Study

TL;DR

A disconnect between image enhancement for human visual perception and for machine analysis is suggested, indicating a need for LLIE methods tailored to support high-level vision tasks effectively.

Abstract

Low-light images are commonly encountered in real-world scenarios, and numerous low-light image enhancement (LLIE) methods have been proposed to improve the visibility of these images. The primary goal of LLIE is to generate clearer images that are more visually pleasing to humans. However, the impact of LLIE methods in high-level vision tasks, such as image classification and object detection, which rely on high-quality image datasets, is not well {explored}. To explore the impact, we comprehensively evaluate LLIE methods on these high-level vision tasks by utilizing an empirical investigation comprising image classification and object detection experiments. The evaluation reveals a dichotomy: {\textit{While Low-Light Image Enhancement (LLIE) methods enhance human visual interpretation, their effect on computer vision tasks is inconsistent and can sometimes be harmful. }} Our findings suggest a disconnect between image enhancement for human visual perception and for machine analysis, indicating a need for LLIE methods tailored to support high-level vision tasks effectively. This insight is crucial for the development of LLIE techniques that align with the needs of both human and machine vision.

Figures (8)

• Figure 1: Object detection examples of YOLOv7 yolov7 on ExDark Exdark and light-enhanced ExDark (using KinD KinD and Zero Zero) dataset.
• Figure 2: Visual multi-level illumination of our low-light synthesis.
• Figure 3: Illuminance information's statistics of the proposed Caltech256-LL and Dark Class dataset.
• Figure 4: Example images from the proposed Dark Class dataset. 'Bike-M' means mountain bike.
• Figure 5: (a) and (b) represent quantitative comparisons of image classification on the Caltech256-LL and Dark Class datasets. (c) denotes quantitative comparisons of the image classification trained on the Caltech and tested on the Caltech-LL dataset. d) represent quantitative comparisons of object detection on the ExDark dataset.