Guardado en:
Detalles Bibliográficos
Autores principales: Zhou, Fuheng, Wei, Dikai, Fan, Ye, Huang, Yulong, Zhang, Yonggang
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2405.16197
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917674935648256
author Zhou, Fuheng
Wei, Dikai
Fan, Ye
Huang, Yulong
Zhang, Yonggang
author_facet Zhou, Fuheng
Wei, Dikai
Fan, Ye
Huang, Yulong
Zhang, Yonggang
contents Although deep learning based models for underwater image enhancement have achieved good performance, they face limitations in both lightweight and effectiveness, which prevents their deployment and application on resource-constrained platforms. Moreover, most existing deep learning based models use data compression to get high-level semantic information in latent space instead of using the original information. Therefore, they require decoder blocks to generate the details of the output. This requires additional computational cost. In this paper, a lightweight network named lightweight selective attention network (LSNet) based on the top-k selective attention and transmission maps mechanism is proposed. The proposed model achieves a PSNR of 97\% with only 7K parameters compared to a similar attention-based model. Extensive experiments show that the proposed LSNet achieves excellent performance in state-of-the-art models with significantly fewer parameters and computational resources. The code is available at https://github.com/FuhengZhou/LSNet}{https://github.com/FuhengZhou/LSNet.
format Preprint
id arxiv_https___arxiv_org_abs_2405_16197
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A 7K Parameter Model for Underwater Image Enhancement based on Transmission Map Prior
Zhou, Fuheng
Wei, Dikai
Fan, Ye
Huang, Yulong
Zhang, Yonggang
Computer Vision and Pattern Recognition
Image and Video Processing
Although deep learning based models for underwater image enhancement have achieved good performance, they face limitations in both lightweight and effectiveness, which prevents their deployment and application on resource-constrained platforms. Moreover, most existing deep learning based models use data compression to get high-level semantic information in latent space instead of using the original information. Therefore, they require decoder blocks to generate the details of the output. This requires additional computational cost. In this paper, a lightweight network named lightweight selective attention network (LSNet) based on the top-k selective attention and transmission maps mechanism is proposed. The proposed model achieves a PSNR of 97\% with only 7K parameters compared to a similar attention-based model. Extensive experiments show that the proposed LSNet achieves excellent performance in state-of-the-art models with significantly fewer parameters and computational resources. The code is available at https://github.com/FuhengZhou/LSNet}{https://github.com/FuhengZhou/LSNet.
title A 7K Parameter Model for Underwater Image Enhancement based on Transmission Map Prior
topic Computer Vision and Pattern Recognition
Image and Video Processing
url https://arxiv.org/abs/2405.16197