Nighttime low-light image enhancement and object detection based on knowledge distillation
Keywords:
deep learning, knowledge distillationAbstract
In order to enhance the quality of nighttime low-light image, improve the accuracy of the object detection model under the nighttime low-light condition and reduce the calculation cost of the model, a multitask model for nighttime low-light image enhancement and object detection based on knowledge distillation and data enhancement was proposed. Knowledge distillation was performed based on the high-quality image model, and the feature information of the high-quality image was used to guide the model training, so that the model could extract the feature information similar to that of the high-quality image in the nighttime low-light images. These feature information could be used to achieve enhancement of image contrast, denoising and objects detection. The experimental results show that the proposed distillation method can improve the object detection accuracy of nighttime low-light by 16.58%, and the image enhanced by this method can achieve the effect of mainstream image enhancement based on deep learning. © 2023 Editorial office of Journal of Applied Optics. All rights reserved.
References
GIRSHICK R, DONAHUE J, DARRELL T, Et al., Rich feature hierarchies for accurate object detection and semantic segmentation, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 580-587, (2014); REDMON J, DIVVALA S, GIRSHICK R, Et al., You only look once: unified, real-time object detection, Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 779-788, (2016); LIU W, ANGUELOV D, ERHAN D, Et al., SSD: single shot multibox detector, European Conference on Computer Vision, pp. 21-37, (2016); LORE K G, AKINTAYO A, SARKAR S., LLNet: a deep autoencoder approach to natural low-light image enhancement[J], Pattern Recognition, 61, pp. 650-662, (2017); LV F, LU F, WU J, Et al., MBLLEN: low-light image/video enhancement using CNNs[J], Baghdad Monitoring and Verification Center, 220, 1, (2018); WEI C, WANG W, YANG W, Et al., Deep retinex decomposition for low-light enhancement; LI C, GUO J, PORIKLI F, Et al., LightenNet: a convolutional neural network for weakly illuminated image enhancement[J], Pattern Recognition Letters, 104, pp. 15-22, (2018); CHEN C, CHEN Q, XU J, Et al., Learning to see in the dark, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 3291-3300, (2018); XIONG W, LIU D, SHEN X, Et al., Unsupervised real-world low-light image enhancement with decoupled networks; JIANG Y, GONG X, LIU D, Et al., Enlightengan: deep light enhancement without paired supervision[J], IEEE Transactions on Image Processing, 30, pp. 2340-2349, (2021); GUO C, LI C, GUO J, Et al., Zero-reference deep curve estimation for low-light image enhancement, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 1780-1789, (2020); KRUTHIVENTI S S S, SAHAY P, BISWAL R., Low-light pedestrian detection from RGB images using multimodal knowledge distillation, 2017 IEEE International Conference on Image Processing （ICIP）, pp. 4207-4211, (2017); QI L, KUEN J, GU J, Et al., Multi-scale aligned distillation for low-resolution detection, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 14443-14453, (2021); HINTON G, VINYALS O, DEAN J., Distilling the knowledge in a neural network[EB/OL]; GE Z, LIU S, WANG F, Et al., Yolox: exceeding yolo series in 2021; RONNEBERGER O, FISCHER P, BROX T., U-net: convolutional networks for biomedical image segmentation [C], International Conference on Medical image computing and computer-assisted intervention, pp. 234-241, (2015); HE K, ZHANG X, REN S, Et al., Spatial pyramid pooling in deep convolutional networks for visual recognition[J], IEEE Transactions on Pattern Analysis and Machine Intelligence, 37, 9, pp. 1904-1916, (2015); WANG W, XIE E, SONG X, Et al., Efficient and accurate arbitrary-shaped text detection with pixel aggregation network, Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 8440-8449, (2019); ZHU J Y, PARK T, ISOLA P, Et al., Unpaired image-to-image translation using cycle-consistent adversarial networks, Proceedings of the IEEE international conference on computer vision, pp. 2223-2232, (2017); DENG J, DONG W, SOCHER R, Et al., Imagenet: a large-scale hierarchical image database, 2009 IEEE conference on computer vision and pattern recognition, pp. 248-255, (2009); MITTAL A, SOUNDARARAJAN R, BOVIK A C., Making a 'completely blind' image quality analyzer[J], IEEE Signal Processing Letters, 20, 3, pp. 209-212, (2012); LOH Y P, CHAN C S., Getting to know low-light images with the exclusively dark dataset[J], Computer Vision and Image Understanding, 178, pp. 30-42, (2019)
Published
How to Cite
Issue
Section
License
Copyright (c) 2023 Journal of Applied Optics
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
The CC Attribution-NonCommercial 4.0 License allows sharing and adapting the work, provided the creator is credited and the work is not used commercially. Modifications must be indicated, and derivative works under the same license are allowed.
