DRACO-DehazeNet: An Efficient Image Dehazing Network Combining Detail Recovery and a Novel Contrastive Learning Paradigm

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lee, Gao Yu, Dam, Tanmoy, Ferdaus, Md Meftahul, Poenar, Daniel Puiu, Duong, Vu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910860672237568
author Lee, Gao Yu
Dam, Tanmoy
Ferdaus, Md Meftahul
Poenar, Daniel Puiu
Duong, Vu
author_facet Lee, Gao Yu
Dam, Tanmoy
Ferdaus, Md Meftahul
Poenar, Daniel Puiu
Duong, Vu
contents Image dehazing is crucial for clarifying images obscured by haze or fog, but current learning-based approaches is dependent on large volumes of training data and hence consumed significant computational power. Additionally, their performance is often inadequate under non-uniform or heavy haze. To address these challenges, we developed the Detail Recovery And Contrastive DehazeNet, which facilitates efficient and effective dehazing via a dense dilated inverted residual block and an attention-based detail recovery network that tailors enhancements to specific dehazed scene contexts. A major innovation is its ability to train effectively with limited data, achieved through a novel quadruplet loss-based contrastive dehazing paradigm. This approach distinctly separates hazy and clear image features while also distinguish lower-quality and higher-quality dehazed images obtained from each sub-modules of our network, thereby refining the dehazing process to a larger extent. Extensive tests on a variety of benchmarked haze datasets demonstrated the superiority of our approach. The code repository for this work is available at https://github.com/GreedYLearner1146/DRACO-DehazeNet.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14595
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle DRACO-DehazeNet: An Efficient Image Dehazing Network Combining Detail Recovery and a Novel Contrastive Learning Paradigm
Lee, Gao Yu
Dam, Tanmoy
Ferdaus, Md Meftahul
Poenar, Daniel Puiu
Duong, Vu
Computer Vision and Pattern Recognition
Image dehazing is crucial for clarifying images obscured by haze or fog, but current learning-based approaches is dependent on large volumes of training data and hence consumed significant computational power. Additionally, their performance is often inadequate under non-uniform or heavy haze. To address these challenges, we developed the Detail Recovery And Contrastive DehazeNet, which facilitates efficient and effective dehazing via a dense dilated inverted residual block and an attention-based detail recovery network that tailors enhancements to specific dehazed scene contexts. A major innovation is its ability to train effectively with limited data, achieved through a novel quadruplet loss-based contrastive dehazing paradigm. This approach distinctly separates hazy and clear image features while also distinguish lower-quality and higher-quality dehazed images obtained from each sub-modules of our network, thereby refining the dehazing process to a larger extent. Extensive tests on a variety of benchmarked haze datasets demonstrated the superiority of our approach. The code repository for this work is available at https://github.com/GreedYLearner1146/DRACO-DehazeNet.
title DRACO-DehazeNet: An Efficient Image Dehazing Network Combining Detail Recovery and a Novel Contrastive Learning Paradigm
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2410.14595