9.代码 GitHub - GlassyWu/AECR-Net: Contrastive Learning for Compact Single Image Dehazing, CVPR2021
论文:Contrastive Learning for Compact Single Image Dehazing, CVPR 2021代码:github.com/GlassyWu/AEC 1、背景 当前方法的问题:1)使用clear images结合L1/L2重建损失进行训练,容易引起结果偏色;2)侧重于加大网络的深度和宽度,导致计算开销巨大。 为了解决当前问题,作者提出了一种新的 contrastive loss。有雾图像,网...
然而,现有的去雾网络通常具有较大的参数量,这限制了它们在实际应用中的性能和效率。为了解决这个问题,本文提出了一种基于对比学习的紧凑去雾网络(Contrastive Learning for Compact Single Image Dehazing)。该方法的核心思想是利用对比正则化来提高去雾网络的性能。对比正则化是一种有效的正则化技术,它通过比较模糊图像...
(1)Adaptive Mixup for Feature Preserving:动态融合下采样层和上采样层之间的特征以实现特征保留; Adaptive mixup dynamically fuses the features between the downsampling layers and the upsampling layers for feature preserving; (2)Dynamic Feature Enhancement(DFE):通过融合更具有空间结构的信息来增强转换能力。
Fork49 Star349 main BranchesTags Code Folders and files Name Last commit message Last commit date Latest commit History 13 Commits .idea data_utils img models README.md metrics.py option.py train_aecrnet.py AECR-Net Contrastive Learning for Compact Single Image Dehazing, CVPR2021. Official Pytor...
Heterogeneous Contrastive Learning: Encoding Spatial Information for Compact Visual Representationsdoi:10.1109/TMM.2021.3115335Huo XinyueLingxi XieLonghui WeiXiaopeng ZhangChen XinLi HaoYang ZijieWengang ZhouHouqiang LiQi TianInstitute of Electrical and Electronics Engineers (IEEE)...
The implementation of NeurIPS 2022 paperExpectation-Maximization Contrastive Learning for Compact Video-and-Language Representations. 💡 I also have other video-language projects that may interest you ✨. Video-Text as Game Players: Hierarchical Banzhaf Interaction for Cross-Modal Representation Learning...
This is the MindSpore version ofContrastive Learning for Compact Single Image Dehazing, CVPR2021 The official PyTorch implementation, pretrained models and examples are available athttps://github.com/GlassyWu/AECR-Net The DCN_v2 module is based onhttps://gitee.com/mindspore/mindspore/tree/master/mod...
几篇论文实现代码:《Contrastive Learning for Many-to-many Multilingual Neural Machine Translation》(ACL 2021) GitHub:https:// github.com/PANXiao1994/mRASP2 [fig2] 《Unifying Nonlocal Blocks for Ne...
论文:Contrastive Learning for Compact Single Image Dehazing 代码:github.com/GlassyWu/AEC 背景: 由于严重的信息退化,单个图像去模糊处理是一个具有挑战性的不适定问题。然而,现有的基于深度学习的去雾方法主要存在两个问题: 仅对正向去雾目标函数的有效性较低。只利用清晰的图像作为正样本,以指导去雾网络的训练...