Spatial

论文解读:医学影像中的注意力机制

烂漫一生 提交于 2020-08-07 16:49:47
点击上方 “ AI小白学视觉 ”,选择加" 星标 "或“置顶” 重磅干货,第一时间送达 来源|Daniel Liu@知乎,https://zhuanlan.zhihu.com/p/138555896 Multi-scale self-guided attention for medical image segmentation 该论文的方法为2019年发布的在CHAOS MRI Dataset上进行医学图像分割的最优方法,最终的Dice为86.75. Introduction:对过往医学图像分割方法的看法 一、对于过往的模型的看法 1、在多数经典模型中对于多尺寸的使用,如unet结构,FCN结构。由于在一开始就是 同样 的low-level信息进行不断的特征提取,所以会造成信息的 冗余 使用。 2、过往模型应用在像素级别的分割挑战中的时候(如,医学领域的分割),可能会表现出判别能力的不足。 二、对于当下用于提高学习特征表达能力的方法,如多尺度的上下文融合,使用空洞卷积,pooling等方式的看法 1、尽管之前的做法可以获得目标在不同尺寸下的信息,但是对于所有的image的上下文联系都是homogenous的和非自适应的, 忽略 了在不同类别中,local-feature和上下文依赖之间的差异。 2、这些多尺度的上下文依赖基本上都是人为设定的,缺乏模型自身的灵活性.

从经典到最新前沿,一文概览2D人体姿态估计

烈酒焚心 提交于 2020-08-04 19:32:40
点击上方“ 3D视觉工坊 ”,选择“星标” 干货第一时间送达 作者:谢一宾 | 来源:知乎 https://zhuanlan.zhihu.com/p/140060196 本文仅做学术分享,如有侵权,请联系删除。 前言 本文主要讨论2D的人体姿态估计,内容主要包括:基本任务介绍、存在的主要困难、方法以及个人对这个问题的思考等等。希望大家带着批判的目光阅读这篇文章,和谐讨论。 介绍 2D人体姿态估计的目标是定位并识别出人体关键点,这些关键点按照关节顺序相连,就可以得到人体的躯干,也就得到了人体的姿态。 在深度学习时代之前,和其他计算机视觉任务一样,都是借助于精心设计的特征来处理这个问题的,比如pictorial structure。凭借着CNN强大的特征提取能力,姿态估计这个领域得到了长足的发展。2D人体姿态估计主要可以分为单人姿态估计(Single Person Pose Estimation, SPPE)和多人姿态估计(Multi-person Pose Estimation, MPPE)两个子任务。 单人姿态估计是基础,在这个问题中,我们要做的事情就是给我们一个人的图片,我们要找出这个人的所有关键点,常用的MPII数据集就是单人姿态估计的数据集。 在多人姿态估计中,我们得到的是一张多人的图,我们需要找出这张图中的所有人的关键点。对于这个问题,一般有自上而下(Top-down

DCGAN论文导读、关键点说明及代码实现修改(1)

血红的双手。 提交于 2020-08-04 10:16:57
论文导读 网上关于DCGAN论文的介绍很多,我就把我觉得对于需要理解的关键点和对后面训练调参有帮助的地方拿出来说明一下,仅做参考,有错误希望大佬们指正。 0.Abstract In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning 这里作者说希望缩小CNN在有监督学习和无监督学习之间应用的差距,也就是当前CNN在有监督场景下应用效果更好。这里关于GAN属于无监督学习我们在后面的训练部分会有深入说明。 1.Introduction We propose that one way to build good image representations is by training Generative Adversarial Networks (GANs) (Goodfellow et al., 2014), and later reusing parts of the generator and discriminator networks as feature extractors for supervised tasks 这里作者说通过训练对抗生成神经网络构建了一种更好的图像表示的方法(one

使用GeoTools 调用WFS服务(Java)

让人想犯罪 __ 提交于 2020-07-27 08:23:43
/** Title: TestBufferAnalysis.java Description: Copyright: Copyright (c) 2020 */ package com.shuidao01.test.geotools; import java.io.IOException; import java.util.HashMap; import java.util.Map; import org.geotools.data.DataStore; import org.geotools.data.DataStoreFinder; import org.geotools.data.FeatureSource; import org.geotools.data.Query; import org.geotools.factory.CommonFactoryFinder; import org.geotools.feature.FeatureCollection; import org.geotools.feature.FeatureIterator; import org.geotools.util.factory.GeoTools; import org.locationtech.jts.geom.Coordinate; import org.locationtech.jts

ArcGIS Pro空间查询

陌路散爱 提交于 2020-07-25 12:26:15
using (Geodatabase geodatabase = new Geodatabase(connectionProperties)) using (FeatureClass schoolBoundaryFeatureClass = geodatabase.OpenDataset<FeatureClass>( " LocalGovernment.GDB.SchoolBoundary " )) { // Using a spatial query filter to find all features which have a certain district name and lying within a given Polygon. SpatialQueryFilter spatialQueryFilter = new SpatialQueryFilter { WhereClause = " DISTRCTNAME = 'Indian Prairie School District 204' " , FilterGeometry = new PolygonBuilder( new List<Coordinate2D> { new Coordinate2D( 1021880 , 1867396 ), new Coordinate2D( 1028223 , 1870705 )

MapBox ESRI Data Layer

谁说胖子不能爱 提交于 2020-07-08 02:44:01
问题 I have a mapbox, and want to display a layer of esri data onto it. The data I'm getting is being pulled in from this json file: https://gis.usps.com/arcgis/rest/services/EDDM/selectZIP/GPServer/routes/execute?f=json&env%3AoutSR=102100&ZIP=93003&Rte_Box=R&UserName=EDDM The data['results'][0]['value']['features'] array looks something like this: [{'attributes': {'key':'value'}},{'geometry':{'paths':[[-13273770,4064608],[-13273762,4064613],....]}}, {'attributes': {'key':'value'}},{'geometry':{

MapBox ESRI Data Layer

谁说胖子不能爱 提交于 2020-07-08 02:43:19
问题 I have a mapbox, and want to display a layer of esri data onto it. The data I'm getting is being pulled in from this json file: https://gis.usps.com/arcgis/rest/services/EDDM/selectZIP/GPServer/routes/execute?f=json&env%3AoutSR=102100&ZIP=93003&Rte_Box=R&UserName=EDDM The data['results'][0]['value']['features'] array looks something like this: [{'attributes': {'key':'value'}},{'geometry':{'paths':[[-13273770,4064608],[-13273762,4064613],....]}}, {'attributes': {'key':'value'}},{'geometry':{

data frame object to use as as.linnet object

人走茶凉 提交于 2020-06-17 07:28:29
问题 I have a data.frame object which I can easily convert to a spatialpointdataframe then convert that to a spatiallinesdataframe but then when I tried to cover to a as.linnet it does not read marks X Y roadID 1 177321.3 3378163 1 2 177321.4 3378168 1 3 177321.4 3378168 1 4 177321.5 3378177 1 5 177321.5 3378186 1 6 177321.5 3378195 1 then I make this data.frame to a SpatialPointsDataFrame coordinates(roaDF1) <- c("X","Y") proj4string(roaDF1)=proj4string(trtrtt) class : SpatialPointsDataFrame

data frame object to use as as.linnet object

不想你离开。 提交于 2020-06-17 07:28:10
问题 I have a data.frame object which I can easily convert to a spatialpointdataframe then convert that to a spatiallinesdataframe but then when I tried to cover to a as.linnet it does not read marks X Y roadID 1 177321.3 3378163 1 2 177321.4 3378168 1 3 177321.4 3378168 1 4 177321.5 3378177 1 5 177321.5 3378186 1 6 177321.5 3378195 1 then I make this data.frame to a SpatialPointsDataFrame coordinates(roaDF1) <- c("X","Y") proj4string(roaDF1)=proj4string(trtrtt) class : SpatialPointsDataFrame

MyDLNote

回眸只為那壹抹淺笑 提交于 2020-05-09 16:11:36
MyDLNote - Attention: [2020 CVPR] Exploring Self-attention for Image Recognition [PAPER] Exploring Self-attention for Image Recognition 目录 MyDLNote - Attention: [2020 CVPR] Exploring Self-attention for Image Recognition Abstract Introduction Related Work Self-attention Networks Pairwise Self-attention Patchwise Self-attention Self-attention Block Comparison Abstract Recent work has shown that self-attention can serve as a basic building block for image recognition models. We explore variations of self-attention and assess their effectiveness for image recognition. We consider two forms of self