已入深夜,您辛苦了!由于当前在线用户较少,发布求助请尽量完整的填写文献信息,科研通机器人24小时在线,伴您度过漫漫科研夜!祝你早点完成任务,早点休息,好梦!

More than Encoder: Introducing Transformer Decoder to Upsample

增采样 分割 计算机科学 人工智能 像素 图像分割 编码器 计算机视觉 杠杆(统计) 模式识别(心理学) 图像(数学) 操作系统
作者
Yijiang Li,Wentian Cai,Ying Gao,Xiping Hu
出处
期刊:Cornell University - arXiv 被引量:1
标识
DOI:10.48550/arxiv.2106.10637
摘要

Medical image segmentation methods downsample images for feature extraction and then upsample them to restore resolution for pixel-level predictions. In such a schema, upsample technique is vital in restoring information for better performance. However, existing upsample techniques leverage little information from downsampling paths. The local and detailed feature from the shallower layer such as boundary and tissue texture is particularly more important in medical segmentation compared with natural image segmentation. To this end, we propose a novel upsample approach for medical image segmentation, Window Attention Upsample (WAU), which upsamples features conditioned on local and detailed features from downsampling path in local windows by introducing attention decoders of Transformer. WAU could serve as a general upsample method and be incorporated into any segmentation model that possesses lateral connections. We first propose the Attention Upsample which consists of Attention Decoder (AD) and bilinear upsample. AD leverages pixel-level attention to model long-range dependency and global information for a better upsample. Bilinear upsample is introduced as the residual connection to complement the upsampled features. Moreover, considering the extensive memory and computation cost of pixel-level attention, we further design a window attention scheme to restrict attention computation in local windows instead of the global range. We evaluate our method (WAU) on classic U-Net structure with lateral connections and achieve state-of-the-art performance on Synapse multi-organ segmentation, Medical Segmentation Decathlon (MSD) Brain, and Automatic Cardiac Diagnosis Challenge (ACDC) datasets. We also validate the effectiveness of our method on multiple classic architectures and achieve consistent improvement.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
2秒前
3秒前
oleskarabach完成签到,获得积分20
5秒前
7秒前
三度和弦发布了新的文献求助10
7秒前
木有完成签到 ,获得积分10
10秒前
JYH12138发布了新的文献求助10
10秒前
ZSQ完成签到 ,获得积分10
11秒前
costahe发布了新的文献求助30
11秒前
Owen应助三度和弦采纳,获得10
12秒前
cheyy发布了新的文献求助10
14秒前
15秒前
JYH12138完成签到,获得积分10
15秒前
拥有八根情丝完成签到 ,获得积分10
16秒前
研友_VZG7GZ应助JYH12138采纳,获得10
18秒前
科研通AI2S应助ReX547413采纳,获得10
19秒前
caitlin完成签到 ,获得积分10
19秒前
陈伟杰发布了新的文献求助10
21秒前
菲1208完成签到,获得积分10
22秒前
23秒前
costahe完成签到,获得积分10
23秒前
gc完成签到 ,获得积分10
23秒前
激动的晓筠完成签到 ,获得积分10
24秒前
Orange应助科研通管家采纳,获得10
24秒前
搜集达人应助科研通管家采纳,获得10
24秒前
脑洞疼应助喂喂采纳,获得10
24秒前
24秒前
三度和弦发布了新的文献求助10
28秒前
阿菜完成签到,获得积分10
28秒前
田様应助陈伟杰采纳,获得10
29秒前
orixero应助三度和弦采纳,获得10
32秒前
文艺的枫叶完成签到 ,获得积分10
32秒前
托丽莲睡拿完成签到,获得积分10
34秒前
leileilei完成签到,获得积分10
35秒前
紫菜汤完成签到 ,获得积分10
35秒前
李健应助萤火虫采纳,获得10
37秒前
希望天下0贩的0应助灵鹿采纳,获得10
38秒前
310769994完成签到,获得积分10
39秒前
40秒前
45秒前
高分求助中
Continuum Thermodynamics and Material Modelling 3000
Production Logging: Theoretical and Interpretive Elements 2700
Mechanistic Modeling of Gas-Liquid Two-Phase Flow in Pipes 2500
Kelsen’s Legacy: Legal Normativity, International Law and Democracy 1000
Conference Record, IAS Annual Meeting 1977 610
Interest Rate Modeling. Volume 3: Products and Risk Management 600
Interest Rate Modeling. Volume 2: Term Structure Models 600
热门求助领域 (近24小时)
化学 材料科学 生物 医学 工程类 有机化学 生物化学 物理 纳米技术 计算机科学 内科学 化学工程 复合材料 基因 遗传学 物理化学 催化作用 量子力学 光电子学 冶金
热门帖子
关注 科研通微信公众号,转发送积分 3544330
求助须知:如何正确求助?哪些是违规求助? 3121530
关于积分的说明 9347654
捐赠科研通 2819788
什么是DOI,文献DOI怎么找? 1550415
邀请新用户注册赠送积分活动 722526
科研通“疑难数据库(出版商)”最低求助积分说明 713265