Posts

#DeepLab# 中的空洞卷积和条件随机场 简要总结 DeepLab V1 和 DeepLab V2，DeepLab 是语义分割的经典网络架构。其有两个核心要点： （1）空洞卷积（Atrous Con...

Boundary loss for highly unbalanced segmentation Csurka, G., Larlus, D., & Perronnin, F. (2013). What is a good evaluation measure for semantic segmentation? BMVC. (218 Citations.) Kervadec, H., Bouchtiba, J., Desrosiers, C., Granger, É., Dolz, J., & Ayed, I.B. (2019). Boundary loss for highly unbalanced segmentation. Medical image analysis, 67, 101851 . (95 Citations) 论文 PDF : https://openreview.net/pdf?id=S1gTA5VggE 或者（https:...

视网膜血管分割 Retinal vessel segmentation 1. 数据集（Datasets） 视网膜血管分割的公开集最常用的有 DRIVE、STARE 和 CHASE_DB。 1.1 DRIVE DRIVE: Digital Retinal Images for Vessel Extraction...

!ls Histogram and RoI.ipynb axon02.tif test.jpg Question 2.1 .ipynb cell_nucleus.tif axon01.tif roi.tif from PIL import Image import numpy as np from PIL import ImageOps import matplotlib.pyplot as plt # 打开图片并转换成灰度图 im = Image.open('axon01.tif').convert('L') # 反转颜色 im = ImageOps.invert(im) # display in macOS /Applications/Preview.app # im.show() # display in Jupyter Notebook im ​ ​ # 将灰度...

MNIST (Mixed National Institute of Standards and Technology database) 参考链接： http://yann.lecun.com/exdb/mnist/ https://stackoverflow.com/questions/40427435/extract-images-from-idx3-ubyte-file-or-gzip-via-python 下载 import glob path = glob.glob('./../data/MNIST/raw/*.gz') path ['./../data/MNIST/raw/t10k-images-idx3-ubyte.gz', './../data/MNIST/raw/train-images-idx3-ubyte.gz', './../data/MNIST/raw/train-labels-idx1-ubyte.gz', './../data/MNIST/raw/t10k-labels-idx1-ubyte.gz'] # train-images-idx3-ubyte.gz # 60000张训练集图片 # train-labels-idx1-ubyte.gz # 60000张训练集图片对应的标签 # t10k-images-idx3-ubyte.gz # 10000张...

数据预处理 import matplotlib.pyplot as plt import glob from PIL import Image import numpy as np import cv2 # 训练集图像 img_paths = glob.glob('./../data/vesselseg/DRIVE/training/images/*.tif') img_paths.sort() img_paths ['./../data/vesselseg/DRIVE/training/images/21_training.tif', './../data/vesselseg/DRIVE/training/images/22_training.tif', './../data/vesselseg/DRIVE/training/images/23_training.tif', './../data/vesselseg/DRIVE/training/images/24_training.tif', './../data/vesselseg/DRIVE/training/images/25_training.tif', './../data/vesselseg/DRIVE/training/images/26_training.tif', './../data/vesselseg/DRIVE/training/images/27_training.tif', './../data/vesselseg/DRIVE/training/images/28_training.tif', './../data/vesselseg/DRIVE/training/images/29_training.tif', './../data/vesselseg/DRIVE/training/images/30_training.tif', './../data/vesselseg/DRIVE/training/images/31_training.tif', './../data/vesselseg/DRIVE/training/images/32_training.tif', './../data/vesselseg/DRIVE/training/images/33_training.tif', './../data/vesselseg/DRIVE/training/images/34_training.tif', './../data/vesselseg/DRIVE/training/images/35_training.tif', './../data/vesselseg/DRIVE/training/images/36_training.tif', './../data/vesselseg/DRIVE/training/images/37_training.tif', './../data/vesselseg/DRIVE/training/images/38_training.tif', './../data/vesselseg/DRIVE/training/images/39_training.tif', './../data/vesselseg/DRIVE/training/images/40_training.tif'] 查看不同通道 # 读取其中一张图像 rgb_img = Image.open(img_paths[0])...

FCN：Fully Convolutional Networks for Semantic Segmentation https://arxiv.org/abs/1411.4038 U-Net：U-Net: Convolutional Networks for Biomedical Image Segmentation https://arxiv.org/abs/1505.04597 SegNet：SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation https://arxiv.org/pdf/1511.00561.pdf 概览 论文所要解决的问题： encoder...

SegNet http://mi.eng.cam.ac.uk/projects/segnet/ 时间：2015 年 Max Pooling 后特征图会变小，空间分辨率的下降对于边界的勾画是不利的，在文章的论述中，作者认为需要对这种边界信息进行存储保留，比如...

Image Segmentation Using Deep Learning: A Survey 一篇图像分割的综述。 arXiv：https://arxiv.org/abs/2001.05566 写这篇笔记时（2021.04....

arXiv: https://arxiv.org/abs/2004.08955 时间：2020 年 04 月 SENet（17.09）、SKNet（19.03） 和 ResNeSt（20.04） 可以放在一起进行阅读，这三篇都是注意...