LOL-YOLO：融合多注意力机制的低照度目标检测

doi:10.3778/j.issn.1002-8331.2406-0424

摘要/Abstract

摘要： 针对低照度图像中目标检测面临的夜间模糊场景、边界不明显场景、明暗差异较大场景等挑战，提出了一种动态特征融合的的检测方法LOL-YOLO（low-light YOLO）。引入了自校正照明模块改善低光照图片的质量，应对低照度下的目标不明显问题；提出了动态特征提取模块，采用结合了大卷积核和可变形卷积的注意力机制，广泛灵活的捕捉图像的上下文信息；设计动态检测头增强对不同尺度、空间位置和任务的感知能力，进一步提升目标检测的精度和鲁棒性。采用ExDark、DarkFace、NPD（nighttime pedestrian detection）数据集进行实验验证，实验结果表明，提出的方法与主流算法相比检测精度明显提升，充分验证了该方法的有效性。

关键词: 低照度, 图像增强, 大卷积核, 可变形卷积, 多重注意力机制

Abstract: Addressing the challenges in low-illumination target detection, such as blurry night scenes, indistinct boundaries, and pronounced brightness disparities, this paper introduces LOL-YOLO (low-light YOLO), a detection method based on dynamic feature fusion. A self-correcting illumination module is incorporated to enhance low-light image quality and counteract target obscurity under low illumination. A dynamic feature extraction module is proposed, which leverages an attention mechanism combining large convolutional kernels with deformable convolutions, enabling extensive and agile contextual information capture. Finally, a dynamic detection head is devised to augment perception of varying scales, spatial positions, and tasks, thereby refining detection accuracy and robustness. Experimental validation using the ExDark, DarkFace, and NPD (nighttime pedestrian detection) datasets demonstrate significant accuracy improvements over prevalent algorithms, confirming the effectiveness of the proposed method.

Key words: low-light, image enhancement, large convolutional kernels, deformable convolutions, multiple attention mechanisms

蒋畅江, 何旭颖, 向杰. LOL-YOLO：融合多注意力机制的低照度目标检测[J]. 计算机工程与应用, 2024, 60(24): 177-187.

JIANG Changjiang, HE Xuying, XIANG Jie. LOL-YOLO：Low-Light Object Detection Incorporating Multiple Attention Mechanisms[J]. Computer Engineering and Applications, 2024, 60(24): 177-187.

参考文献

[1] 王满利, 张航, 张长森. 基于深度学习的低光照目标检测算法[J/OL]. 北京邮电大学学报(2023-08-01)[2024-08-07]. https://doi.org/10.13190/j.jbupt.2023-161.
WANG M L, ZHANG H, ZHANG C S. Deep learning based algorithm for low light target detection[J/OL]. Journal of Beijing University of Posts and Telecommunication (2023-08-01)[2024-08-07]. https://doi.org/10.13190/j.jbupt.2023-161.
[2] WANG K, LIU M Z. Object recognition at night scene based on DCGAN and faster R-CNN[J]. IEEE Access, 2020, 8: 193168-193182.
[3] XU Y, CHU K, ZHANG J. Nighttime vehicle detection algorithm based on improved Faster-RCNN[J]. IEEE Access, 2023, 12: 19299-19306.
[4] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision (ECCV 2016), Amsterdam, The Netherlands, October 11-14, 2016: 21-37.
[5] TAN M, PANG R, LE Q V. Efficientdet: scalable and efficient object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 10781-10790.
[6] 王琳毅, 白静, 李文静, 等. YOLO 系列目标检测算法研究进展[J]. 计算机工程与应用, 2023, 59(14): 15-29.
WANG L Y, BAI J, LI W J, et al. Research progress of YOLO series target detection algorithms[J]. Computer Engineering and Applications, 2023, 59(14): 15-29.
[7] SASAGAWA Y, NAGAHARA H. YOLO in the dark-domain adaptation method for merging multiple models[C]//Proceedings of the 16th European Conference on Computer Vision (ECCV 2020), Glasgow, UK, August 23-28, 2020: 345-359.
[8] QIN Q P, CHANG K, HUANG M Y, et al. DENet: detection-driven enhancement network for object detection under adverse weather conditions[C]//Proceedings of the Asian Conference on Computer Vision, 2022: 2813-2829.
[9] JIANG Q, MAO Y, CONG R, et al. Unsupervised decomposition and correction network for low-light image enhancement[J]. IEEE Transactions on Intelligent Transportation Systems, 2022, 23(10): 19440-19455.
[10] HU M, WANG S, LI B, et al. Penet: towards precise and efficient image guided depth completion[C]//Proceedings of the 2021 IEEE International Conference on Robotics and Automation (ICRA), 2021: 13656-13662.
[11] CUI Z, LI K, GU L, et al. You only need 90k parameters to adapt light: a light weight transformer for image enhancement and exposure correction[C]//Proceedings of the 33rd British Machine Vision Conference 2022 (BMVC 2022), London, UK, November 21-24, 2022.
[12] VINOTH K P S. Lightweight object detection in low light: pixel-wise depth refinement and TensorRT optimization[J]. Results in Engineering, 2024, 23: 102510.
[13] YU N N, WANG J, SHI H, et al. Degradation-removed multiscale fusion for low-light salient object detection[J]. Pattern Recognition, 2024, 155: 110650.
[14] YUE H, GUO J, YIN X, et al. Salient object detection in low-light RGB-T scene via spatial-frequency cues mining[J]. Neural Networks: the Official Journal of the International Neural Network Society, 2024, 178: 106406.
[15] JING L, WANG B. EMNet: edge-guided multi-level network for salient object detection in low-light images[J]. Image and Vision Computing, 2024, 143: 104933.
[16] HONG Y, WEI K, CHEN L, et al. Crafting object detection in very low light[C]//Proceedings of the 32nd British Machine Vision Conference, 2021.
[17] MA L, MA T Y, LIU R S, et al. Penet: toward fast, flexible, and robust low-light image enhancement[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, LA, USA, 2022: 5627-5636.
[18] WANG W, DAI J, CHEN Z, et al. Intern image: exploring large-scale vision foundation models with deformable convolutions[C]//Proceedings of the 2023 IEEE Conference on Computer Vision and Pattern Recognition, Vancouver, Canada, 2023: 14408-14419.
[19] DAI X, CHEN Y, XIAO B, et al. Dynamic head: unifying object detection heads with attentions[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 7373-7382.
[20] LIU X, PENG H, ZHENG N, et al. EfficientViT: memory efficient vision transformer with cascaded group attention[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 14420-14430.
[21] WAN D H, LU R S, SHEN S Y, et al. Mixed local channel attention for object detection[J]. Engineering Applications of Artificial Intelligence, 2023, 123: 106442.
[22] MISRA D, NALAMADA T, ARASANIPALAI A U, et al. Rotate to attend: convolutional triplet attention module[C]//Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2021: 3139-3148.
[23] ZHANG J, LI X, LI J, et al. Rethinking mobile block for efficient attention-based models[C]//Proceedings of the 2023 IEEE/CVF International Conference on Computer Vision (ICCV), 2023: 1389-1400.
[24] DING X, ZHANG X, MA N, et al. Repvgg: making VGG-style convnets great again[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 13733-13742.
[25] DING X, ZHANG X, HAN J, et al. Diverse branch block: building a convolution as an inception-like unit[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 10886-10895.
[26] CUI Z T, QI G J, GU L, et al. Multitask AET with orthogonal tangent regularity for dark object detection[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021: 2553-2562.
[27] LIU W Y, REN G F, YU R S, et al. Image-adaptive YOLO for object detection in adverse weather conditions[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2022: 1792-1800.
[28] 谭豪, 张惊雷, 贾鑫. 基于多级特征提取的低光照目标检测算法[J]. 计算机工程与应用2024, 60(24): 235-242.
TAN H, ZHANG J L, JIA X. Low-light target detection algorithm based on multi-level feature extraction[J]. Computer Engineering and Applications, 2024, 60(24): 235-242.
[29] WANG J, YANG P, LIU Y, et al. Research on improved YOLOv5 for low-light environment object detection[J]. Electronics, 2023, 12(14): 3089.
[30] LIU R, MA L, ZHANG J, et al. Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 10561-10570.
[31] GUO C L, LI C Y, GUO J C, et al. Zero-reference deep curve estimation for low-light image enhancement[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 1780-1789.