|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/unet-tensorflow2/
课程评论：没有评论
**ourse summary: UNet (TensorFlow 2) Image Semantic Segmentation Practical: Training Your Own Dataset** This course provides a hands-on guide to image semantic segmentation using the U-Net architecture with TensorFlow 2. U-Net is a deep learning method particularly effective for medical image segmentation. The course will walk you through the entire process of building your own semantic segmentation application, from creating a custom dataset to training and evaluating a U-Net model. **Key topics covered:** * **Introduction to Image Segmentation:** Understanding the task, common datasets, and applications. * **U-Net Architecture:** In-depth explanation of the U-Net's design and principles. * **Dataset Creation:** Using the labelme tool for image annotation and generating mask images. * **U-Net Implementation (TensorFlow 2):** Writing U-Net program files and training your own datasets. * **Project Practices:** * Kaggle Salt Identification Competition: Applying U-Net for salt body recognition. * Pothole Semantic Segmentation: Annotating and segmenting potholes in driving scenes. * Kaggle Nuclei Segmentation Competition: Using U-Net for nuclei segmentation. * **Model Training and Evaluation:** Testing trained models and performance assessment. * **Environment Setup:** Demonstrations on Ubuntu using Jupyter Notebook, with instructions for Windows environment setup. The course provides all necessary project datasets and Python code.
U-Net是一种基于深度学习的图像语义分割方法，尤其在医学图像分割中表现优异。本课程将手把手地教大家使用labelme图像标注工具制作自己的数据集，生成Mask图像，并使用U-Net训练自己的数据集，从而能开展自己的图像分割应用。本课程首先讲述图像分割的任务说明、常用数据集，然后介绍UNet网络的原理。本课程有三个项目实践：(1) Kaggle盐体识别比赛 ：利用U-Net进行Kaggle盐体识别(2) Pothole语义分割：对汽车行驶场景中的路坑进行标注和语义分割(3) Kaggle细胞核分割比赛 ：利用U-Net进行Kaggle细胞核分割本课程使用TensorFlow2版本的U-Net，在Ubuntu系统上用Jupyter Notebook做项目演示。 包括：数据集标注、数据集格式转换和Mask图像生成、编写U-Net程序文件、训练自己的数据集、测试训练出的网络模型、性能评估。项目代码也可在Windows上运行，课程提供Windows环境搭建方法。本课程提供项目的数据集和Python程序代码。
