p47068915
/
mindspore
forked from mindspore-Ecosystem/mindspore
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

update unet readme

This commit is contained in:
zhaoting 2021-03-10 10:22:08 +08:00
parent 063607de90
commit da91db20aa
2 changed files with 119 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
model_zoo/official/cv/unet/README.md
View File

@ -30,7 +30,11 @@
Unet Medical model for 2D image segmentation. This implementation is as described in the original paper [UNet: Convolutional Networks for Biomedical Image Segmentation](https://arxiv.org/abs/1505.04597). Unet, in the 2015 ISBI cell tracking competition, many of the best are obtained. In this paper, a network model for medical image segmentation is proposed, and a data enhancement method is proposed to effectively use the annotation data to solve the problem of insufficient annotation data in the medical field. A U-shaped network structure is also used to extract the context and location information.
[Paper](https://arxiv.org/abs/1505.04597): Olaf Ronneberger, Philipp Fischer, Thomas Brox. "U-Net: Convolutional Networks for Biomedical Image Segmentation." *conditionally accepted at MICCAI 2015*. 2015.
UNet++ is a neural architecture for semantic and instance segmentation with re-designed skip pathways and deep supervision.
[U-Net Paper](https://arxiv.org/abs/1505.04597): Olaf Ronneberger, Philipp Fischer, Thomas Brox. "U-Net: Convolutional Networks for Biomedical Image Segmentation." *conditionally accepted at MICCAI 2015*. 2015.
[UNet++ Paper](https://arxiv.org/abs/1912.05074): Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh and J. Liang, "UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation," in IEEE Transactions on Medical Imaging, vol. 39, no. 6, pp. 1856-1867, June 2020, doi: 10.1109/TMI.2019.2959609.
## [Model Architecture](#contents)
@ -49,6 +53,8 @@ Dataset used: [ISBI Challenge](http://brainiac2.mit.edu/isbi_challenge/home)
- Data format：binary files(TIF file)
- Note：Data will be processed in src/data_loader.py
We also support cell nuclei dataset which is used in [Unet++ original paper](https://arxiv.org/abs/1912.05074). If you want to use the dataset, please add `'dataset': 'Cell_nuclei'` in `src/config.py`.
## [Environment Requirements](#contents)
- Hardware（Ascend）
@ -63,6 +69,10 @@ Dataset used: [ISBI Challenge](http://brainiac2.mit.edu/isbi_challenge/home)
After installing MindSpore via the official website, you can start training and evaluation as follows:
- Select the network and dataset to use
Refer to `src/config.py`. We support some parameter configurations for quick start. You can set `'model'` to `'unet_medical'`,`'unet_nested'` or `'unet_simple'` to select which net to use. We support `ISBI` and `Cell_nuclei` two dataset, you can set `'dataset'` to `'Cell_nuclei'` to use `Cell_nuclei` dataset, default is `ISBI`.
- Run on Ascend
```python
@ -107,21 +117,32 @@ Then you can run everything just like on ascend.
├── README.md // descriptions about all the models
├── unet
├── README.md // descriptions about Unet
├── ascend310_infer // code of infer on ascend 310
├── scripts
│ ├──run_standalone_train.sh // shell script for distributed on Ascend
│ ├──docker_start.sh // shell script for quick docker start
│ ├──run_disribute_train.sh // shell script for distributed on Ascend
│ ├──run_infer_310.sh // shell script for infer on ascend 310
│ ├──run_standalone_train.sh // shell script for standalone on Ascend
│ ├──run_standalone_eval.sh // shell script for evaluation on Ascend
├── src
│ ├──config.py // parameter configuration
│ ├──data_loader.py // creating dataset
│ ├──loss.py // loss
│ ├──utils.py // General components (callback function)
│ ├──unet.py // Unet architecture
│ ├──unet_medical // Unet medical architecture
├──__init__.py // init file
├──unet_model.py // unet model
├──unet_parts.py // unet part
│ ├──unet_nested // Unet++ architecture
├──__init__.py // init file
├──unet_model.py // unet model
├──unet_parts.py // unet part
├── train.py // training script
├──launch_8p.py // training 8P script
├── eval.py // evaluation script
├── eval.py // evaluation script
├── export.py // export script
├── mindspore_hub_conf.py // hub config file
├── postprocess.py // unet 310 infer postprocess.
├── preprocess.py // unet 310 infer preprocess dataset
```
### [Script Parameters](#contents)
@ -149,6 +170,31 @@ Parameters for both training and evaluation can be set in config.py
'filter_weight': ["final.weight"] # weight name to filter while doing transfer training
```
- config for Unet++, cell nuclei dataset
```python
'model': 'unet_nested', # model name
'dataset': 'Cell_nuclei', # dataset name
'img_size': [96, 96], # image size
'lr': 3e-4, # learning rate
'epochs': 200, # total training epochs when run 1p
'distribute_epochs': 1600, # total training epochs when run 8p
'batchsize': 16, # batch size
'num_classes': 2, # the number of classes in the dataset
'num_channels': 3, # the number of input image channels
'keep_checkpoint_max': 10, # only keep the last keep_checkpoint_max checkpoint
'weight_decay': 0.0005, # weight decay value
'loss_scale': 1024.0, # loss scale
'FixedLossScaleManager': 1024.0, # loss scale
'use_bn': True, # whether to use BN
'use_ds': True, # whether to use deep supervisio
'use_deconv': True, # whether to use Conv2dTranspose
'resume': False, # whether training with pretrain model
'resume_ckpt': './', # pretrain model path
'transfer_training': False # whether do transfer training
'filter_weight': ['final1.weight', 'final2.weight', 'final3.weight', 'final4.weight'] # weight name to filter while doing transfer training
```
## [Training Process](#contents)
### Training
@ -283,10 +329,10 @@ Cross valid dice coeff is: 0.9054352151297033
Set options `resume` to True in `config.py`, and set `resume_ckpt` to the path of your checkpoint. e.g.
```python
'resume': True,
'resume_ckpt': 'ckpt_0/ckpt_unet_medical_adam_1-1_600.ckpt',
'transfer_training': False,
'filter_weight': ["final.weight"]
'resume': True,
'resume_ckpt': 'ckpt_0/ckpt_unet_medical_adam_1-1_600.ckpt',
'transfer_training': False,
'filter_weight': ["final.weight"]
```
#### Transfer training
@ -294,10 +340,10 @@ Set options `resume` to True in `config.py`, and set `resume_ckpt` to the path o
Do the same thing as resuming traing above. In addition, set `transfer_training` to True. The `filter_weight` shows the weights which will be filtered for different dataset. Usually, the default value of `filter_weight` don't need to be changed. The default values includes the weights which depends on the class number. e.g.
```python
'resume': True,
'resume_ckpt': 'ckpt_0/ckpt_unet_medical_adam_1-1_600.ckpt',
'transfer_training': True,
'filter_weight': ["final.weight"]
'resume': True,
'resume_ckpt': 'ckpt_0/ckpt_unet_medical_adam_1-1_600.ckpt',
'transfer_training': True,
'filter_weight': ["final.weight"]
```
## [Description of Random Situation](#contents)

74
model_zoo/official/cv/unet/README_CN.md
View File

@ -33,7 +33,11 @@
U-Net医学模型基于二维图像分割。实现方式见论文[UNet：Convolutional Networks for Biomedical Image Segmentation](https://arxiv.org/abs/1505.04597)。在2015年ISBI细胞跟踪竞赛中，U-Net获得了许多最佳奖项。论文中提出了一种用于医学图像分割的网络模型和数据增强方法，有效利用标注数据来解决医学领域标注数据不足的问题。U型网络结构也用于提取上下文和位置信息。
[论文](https://arxiv.org/abs/1505.04597)： Olaf Ronneberger, Philipp Fischer, Thomas Brox. "U-Net: Convolutional Networks for Biomedical Image Segmentation." *conditionally accepted at MICCAI 2015*. 2015.
UNet++是U-Net的增强版本，使用了新的跨层链接方式和深层监督，可以用于语义分割和实例分割。
[U-Net 论文](https://arxiv.org/abs/1505.04597): Olaf Ronneberger, Philipp Fischer, Thomas Brox. "U-Net: Convolutional Networks for Biomedical Image Segmentation." *conditionally accepted at MICCAI 2015*. 2015.
[UNet++ 论文](https://arxiv.org/abs/1912.05074): Z. Zhou, M. M. R. Siddiquee, N. Tajbakhsh and J. Liang, "UNet++: Redesigning Skip Connections to Exploit Multiscale Features in Image Segmentation," in IEEE Transactions on Medical Imaging, vol. 39, no. 6, pp. 1856-1867, June 2020, doi: 10.1109/TMI.2019.2959609.
## 模型架构
@ -53,6 +57,8 @@ U-Net医学模型基于二维图像分割。实现方式见论文[UNet：Convolu
- 数据格式：二进制文件（TIF）
- 注意：数据在src/data_loader.py中处理
我们也支持一个在 [Unet++](https://arxiv.org/abs/1912.05074) 原论文中使用的数据集 `Cell_nuclei`。可以通过修改`src/config.py`中`'dataset': 'Cell_nuclei'`配置使用.
## 环境要求
- 硬件（Ascend）
@ -67,6 +73,10 @@ U-Net医学模型基于二维图像分割。实现方式见论文[UNet：Convolu
通过官方网站安装MindSpore后，您可以按照如下步骤进行训练和评估：
- 选择模型及数据集
我们在`src/config.py`预备了一些网络及数据集的参数配置用于快速体验。也可以通过设置`'model'` 为 `'unet_medical'`,`'unet_nested'` 或者 `'unet_simple'` 来选择使用什么网络结构。我们支持`ISBI` 和 `Cell_nuclei`两种数据集处理，默认使用`ISBI`，可以设置`'dataset'` 为 `'Cell_nuclei'`使用`Cell_nuclei`数据集。
- Ascend处理器环境运行
```python
@ -108,24 +118,35 @@ bash scripts/docker_start.sh unet:20.1.0 [DATA_DIR] [MODEL_DIR]
```path
├── model_zoo
├── README.md // 所有模型相关说明
├── README.md // 模型描述
├── unet
├── README.md // U-Net相关说明
├── README.md // Unet描述
├── ascend310_infer // Ascend 310 推理代码
├── scripts
│ ├──run_standalone_train.sh // Ascend分布式shell脚本
│ ├──run_standalone_eval.sh // Ascend评估shell脚本
│ ├──docker_start.sh // docker 脚本
│ ├──run_disribute_train.sh // Ascend 上分布式训练脚本
│ ├──run_infer_310.sh // Ascend 310 推理脚本
│ ├──run_standalone_train.sh // Ascend 上单卡训练脚本
│ ├──run_standalone_eval.sh // Ascend 上推理脚本
├── src
│ ├──config.py // 参数配置
│ ├──data_loader.py // 创建数据集
│ ├──loss.py // 损失
│ ├──data_loader.py // 数据处理
│ ├──loss.py // 损失函数
│ ├──utils.py // 通用组件（回调函数）
│ ├──unet.py // U-Net架构
├──__init__.py // 初始化文件
├──unet_model.py // U-Net模型
├──unet_parts.py // U-Net部分
│ ├──unet_medical // 医学图像处理Unet结构
├──__init__.py
├──unet_model.py // Unet 网络结构
├──unet_parts.py // Unet 子网
│ ├──unet_nested // Unet++
├──__init__.py
├──unet_model.py // Unet++ 网络结构
├──unet_parts.py // Unet++ 子网
├── train.py // 训练脚本
├──launch_8p.py // 训练8P脚本
├── eval.py // 评估脚本
├── eval.py // 推理脚本
├── export.py // 导出脚本
├── mindspore_hub_conf.py // hub 配置脚本
├── postprocess.py // 310 推理后处理脚本
├── preprocess.py // 310 推理前处理脚本
```
### 脚本参数
@ -143,7 +164,7 @@ bash scripts/docker_start.sh unet:20.1.0 [DATA_DIR] [MODEL_DIR]
'cross_valid_ind': 1, # 交叉验证指标
'num_classes': 2, # 数据集类数
'num_channels': 1, # 通道数
'keep_checkpoint_max': 10, # 只保留最后一个keep_checkpoint_max检查点
'keep_checkpoint_max': 10, # 保留checkpoint检查个数
'weight_decay': 0.0005, # 权重衰减值
'loss_scale': 1024.0, # 损失放大
'FixedLossScaleManager': 1024.0, # 固定损失放大
@ -151,6 +172,31 @@ bash scripts/docker_start.sh unet:20.1.0 [DATA_DIR] [MODEL_DIR]
'resume_ckpt': './', # 预训练模型路径
```
- Unet++配置, cell nuclei数据集
```python
'model': 'unet_nested', # 模型名称
'dataset': 'Cell_nuclei', # 数据集名称
'img_size': [96, 96], # 输入图像大小
'lr': 3e-4, # 学习率
'epochs': 200, # 运行1p时的总训练轮次
'distribute_epochs': 1600, # 运行8p时的总训练轮次
'batchsize': 16, # 训练批次大小
'num_classes': 2, # 数据集类数
'num_channels': 3, # 输入图像通道数
'keep_checkpoint_max': 10, # 保留checkpoint检查个数
'weight_decay': 0.0005, # 权重衰减值
'loss_scale': 1024.0, # 损失放大
'FixedLossScaleManager': 1024.0, # 损失放大
'use_bn': True, # 是否使用BN
'use_ds': True, # 是否使用深层监督
'use_deconv': True, # 是否使用反卷积
'resume': False, # 是否使用预训练模型训练
'resume_ckpt': './', # 预训练模型路径
'transfer_training': False # 是否使用迁移学习
'filter_weight': ['final1.weight', 'final2.weight', 'final3.weight', 'final4.weight'] # 迁移学习过滤参数名
```
## 训练过程
### 用法