Official implementation of SCTNet (AAAI2024). Contribute to xzz777/SCTNet development by creating an account on GitHub.
Github地址:https://github.com/xzz777/SCTNet SCTNet是一种用于实时语义分割的单分支CNN模型,它利用了Transformer作为训练时的语义分支,以提取长程上下文信息。在推理阶段,只需要部署单分支CNN即可。实验证明,SCTNet在Cityscapes、ADE20K和COCO-Stuff10K数据集上取得了最新的最优性能。 SCTNet的单分支卷积神经网络,用...
Packages No packages published Contributors 3 Zongwei97 yulunzhang Yulun Zhang steven-tel Languages MATLAB 57.6% Python 42.4% Footer © 2025 GitHub, Inc. Footer navigation Terms Privacy Security Status Docs Contact Manage cookies Do not share my personal information ...
然而,额外的分支会带来不必要的计算开销,并减缓推理速度。为了消除这一困境,我们提出了SCTNet,一种带有transformer语义信息的单分支CNN用于实时分割。 https://arxiv.org/abs/2312.17071 https://github.com/xzz777/SCTNet SCTNet在保留轻量级单分支CNN高效性的同时,还拥有语义分支的丰富语义表示。考虑到transformer提取...
https://arxiv.org/abs/2312.17071 https://github.com/xzz777/SCTNet SCTNet在保留轻量级单分支CNN高效性的同时,还拥有语义分支的丰富语义表示。考虑到transformer提取长距离上下文的卓越能力,SCTNet将transformer作为仅用于训练的语义分支。借助于提出的transformer类CNN块CFBlock和语义信息对齐模块,SCTNet可以在训练中从...
Is there anyone who can implement theFADC and SCTNetnetwork codes in mmsegmentation? Many thanks to everyone who has contributed to this. The relevant papers and code repositories for these two networks are as follows: (1)FADC: Code:https://github.com/Linwei-Chen/FADC ...
代码:https://github.com/xzz777/SCTNet 论文下载链接(PDF已上传至星球,可一键下载):https://arxiv.org/abs/2312.17071 最近的实时语义分割方法通常采用额外的语义分支来追求丰富的远程上下文。 然而,额外的分支会产生不需要的计算开销并降低推理速度。
代码和模型可在github.com/xzz777/SCTNe上获得。 1 介绍 语义分割是计算机视觉中的一项基本任务,其目的是为输入图像中的每个像素分配一个语义类标签。它在自动驾驶、医疗图像处理、移动应用和许多其他领域发挥着至关重要的作用。为了获得更好的语义分割性能,最近的语义分割方法追求丰富的远程上下文。已经提出了不同的...
[ECCV 2022 Oral] Official Pytorch implementation of CCPL and SCTNet - GitHub - JarrentWu1031/CCPL: [ECCV 2022 Oral] Official Pytorch implementation of CCPL and SCTNet
Mirrors.NET, the Mirror Manager for ZJU Mirror. Contribute to ZJUSCT/MirrorsDotNet development by creating an account on GitHub.