forked from mindspore-Ecosystem/mindspore
!15305 update psenet readme file
From: @anzhengqi Reviewed-by: @heleiwang,@pandoublefeng Signed-off-by: @pandoublefeng
This commit is contained in:
commit
af8b2d3a47
|
@ -90,31 +90,34 @@ sh scripts/run_eval_ascend.sh
|
||||||
|
|
||||||
```path
|
```path
|
||||||
└── PSENet
|
└── PSENet
|
||||||
├── README.md // descriptions about PSENet
|
├── export.py // export mindir file
|
||||||
|
├── __init__.py
|
||||||
|
├── mindspore_hub_conf.py // hub config file
|
||||||
|
├── README_CN.md // descriptions about PSENet in Chinese
|
||||||
|
├── README.md // descriptions about PSENet in English
|
||||||
├── scripts
|
├── scripts
|
||||||
├── run_distribute_train.sh // shell script for distributed
|
├── run_distribute_train.sh // shell script for distributed
|
||||||
└── run_eval_ascend.sh // shell script for evaluation
|
└── run_eval_ascend.sh // shell script for evaluation
|
||||||
├──src
|
├── src
|
||||||
├── __init__.py
|
├── config.py // parameter configuration
|
||||||
|
├── dataset.py // creating dataset
|
||||||
├── ETSNET
|
├── ETSNET
|
||||||
├── __init__.py
|
|
||||||
├── base.py // convolution and BN operator
|
├── base.py // convolution and BN operator
|
||||||
├── dice_loss.py // calculate PSENet loss value
|
├── dice_loss.py // calculate PSENet loss value
|
||||||
├── etsnet.py // Subnet in PSENet
|
├── etsnet.py // Subnet in PSENet
|
||||||
├── fpn.py // Subnet in PSENet
|
├── fpn.py // Subnet in PSENet
|
||||||
├── resnet50.py // Subnet in PSENet
|
├── __init__.py
|
||||||
├── pse // Subnet in PSENet
|
├── pse // Subnet in PSENet
|
||||||
├── __init__.py
|
├── __init__.py
|
||||||
├── adaptor.cpp
|
├── adaptor.cpp
|
||||||
├── adaptor.h
|
├── adaptor.h
|
||||||
├── Makefile
|
├── Makefile
|
||||||
├──config.py // parameter configuration
|
├── resnet50.py // Subnet in PSENet
|
||||||
├──dataset.py // creating dataset
|
├── __init__.py
|
||||||
├──network_define.py // learning ratio generation
|
├── lr_schedule.py // define learning rate
|
||||||
├──export.py // export mindir file
|
├── network_define.py // learning ratio generation
|
||||||
├──mindspore_hub_conf.py // hub config file
|
├── test.py // test script
|
||||||
├──test.py // test script
|
├── train.py // training script
|
||||||
├──train.py // training script
|
|
||||||
|
|
||||||
```
|
```
|
||||||
|
|
||||||
|
@ -164,7 +167,9 @@ python test.py --ckpt=./device*/ckpt*/ETSNet-*.ckpt
|
||||||
#### Usage
|
#### Usage
|
||||||
|
|
||||||
step 1: download eval method from [here](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization).
|
step 1: download eval method from [here](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization).
|
||||||
|
|
||||||
step 2: click "My Methods" button,then download Evaluation Scripts.
|
step 2: click "My Methods" button,then download Evaluation Scripts.
|
||||||
|
|
||||||
step 3: it is recommended to symlink the eval method root to $MINDSPORE/model_zoo/psenet/eval_ic15/. if your folder structure is different,you may need to change the corresponding paths in eval script files.
|
step 3: it is recommended to symlink the eval method root to $MINDSPORE/model_zoo/psenet/eval_ic15/. if your folder structure is different,you may need to change the corresponding paths in eval script files.
|
||||||
|
|
||||||
```shell
|
```shell
|
||||||
|
@ -181,12 +186,12 @@ Calculated!{"precision": 0.814796668299853, "recall": 0.8006740491092923, "hmean
|
||||||
|
|
||||||
### Evaluation Performance
|
### Evaluation Performance
|
||||||
|
|
||||||
| Parameters | PSENet |
|
| Parameters | Ascend |
|
||||||
| -------------------------- | ----------------------------------------------------------- |
|
| -------------------------- | ----------------------------------------------------------- |
|
||||||
| Model Version | V1 |
|
| Model Version | PSENet |
|
||||||
| Resource | Ascend 910; CPU 2.60GHz, 192cores; Memory 755G; OS Euler2.8 |
|
| Resource | Ascend 910; CPU 2.60GHz, 192cores; Memory 755G; OS Euler2.8 |
|
||||||
| uploaded Date | 09/30/2020 (month/day/year) |
|
| uploaded Date | 09/30/2020 (month/day/year) |
|
||||||
| MindSpore Version | 1.0.0 |
|
| MindSpore Version | 1.0.0 |
|
||||||
| Dataset | ICDAR2015 |
|
| Dataset | ICDAR2015 |
|
||||||
| Training Parameters | start_lr=0.1; lr_scale=0.1 |
|
| Training Parameters | start_lr=0.1; lr_scale=0.1 |
|
||||||
| Optimizer | SGD |
|
| Optimizer | SGD |
|
||||||
|
@ -194,19 +199,19 @@ Calculated!{"precision": 0.814796668299853, "recall": 0.8006740491092923, "hmean
|
||||||
| outputs | probability |
|
| outputs | probability |
|
||||||
| Loss | 0.35 |
|
| Loss | 0.35 |
|
||||||
| Speed | 1pc: 444 ms/step; 8pcs: 446 ms/step |
|
| Speed | 1pc: 444 ms/step; 8pcs: 446 ms/step |
|
||||||
| Total time | 1pc: 75.48 h; 8pcs: 10.01 h |
|
| Total time | 1pc: 75.48 h; 8pcs: 7.11 h |
|
||||||
| Parameters (M) | 27.36 |
|
| Parameters (M) | 27.36 |
|
||||||
| Checkpoint for Fine tuning | 109.44M (.ckpt file) |
|
| Checkpoint for Fine tuning | 109.44M (.ckpt file) |
|
||||||
| Scripts | <https://gitee.com/mindspore/mindspore/tree/master/model_zoo/official/cv/psenet> |
|
| Scripts | <https://gitee.com/mindspore/mindspore/tree/master/model_zoo/official/cv/psenet> |
|
||||||
|
|
||||||
### Inference Performance
|
### Inference Performance
|
||||||
|
|
||||||
| Parameters | PSENet |
|
| Parameters | Ascend |
|
||||||
| ------------------- | --------------------------- |
|
| ------------------- | --------------------------- |
|
||||||
| Model Version | V1 |
|
| Model Version | PSENet |
|
||||||
| Resource | Ascend 910 |
|
| Resource | Ascend 910 |
|
||||||
| Uploaded Date | 09/30/2020 (month/day/year) |
|
| Uploaded Date | 09/30/2020 (month/day/year) |
|
||||||
| MindSpore Version | 1.0,0 |
|
| MindSpore Version | 1.0.0 |
|
||||||
| Dataset | ICDAR2015 |
|
| Dataset | ICDAR2015 |
|
||||||
| outputs | probability |
|
| outputs | probability |
|
||||||
| Accuracy | 1pc: 81%; 8pcs: 81% |
|
| Accuracy | 1pc: 81%; 8pcs: 81% |
|
||||||
|
|
|
@ -60,7 +60,7 @@
|
||||||
|
|
||||||
# 快速入门
|
# 快速入门
|
||||||
|
|
||||||
通过官方网站安装MindSpore后,您可以按照如下步骤进行培训和评估:
|
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:
|
||||||
|
|
||||||
```python
|
```python
|
||||||
# 分布式训练运行示例
|
# 分布式训练运行示例
|
||||||
|
@ -91,31 +91,34 @@ sh scripts/run_eval_ascend.sh
|
||||||
## 脚本和样例代码
|
## 脚本和样例代码
|
||||||
|
|
||||||
```path
|
```path
|
||||||
└── PSENet
|
└── PSENet
|
||||||
├── README.md // PSENet相关描述
|
├── export.py // mindir转换脚本
|
||||||
├── scripts
|
├── mindspore_hub_conf.py // 网络模型
|
||||||
|
├── README.md // PSENet相关描述英文版
|
||||||
|
├── README_CN.md // PSENet相关描述中文版
|
||||||
|
├── scripts
|
||||||
├── run_distribute_train.sh // 用于分布式训练的shell脚本
|
├── run_distribute_train.sh // 用于分布式训练的shell脚本
|
||||||
└── run_eval_ascend.sh // 用于评估的shell脚本
|
└── run_eval_ascend.sh // 用于评估的shell脚本
|
||||||
├──src
|
├── src
|
||||||
├── __init__.py
|
├── config.py // 参数配置
|
||||||
├── generate_hccn_file.py // 创建rank.json文件
|
├── dataset.py // 创建数据集
|
||||||
├── ETSNET
|
├── ETSNET
|
||||||
├── __init__.py
|
|
||||||
├── base.py // 卷积和BN算子
|
├── base.py // 卷积和BN算子
|
||||||
├── dice_loss.py // 计算PSENet损耗值
|
├── dice_loss.py // 计算PSENet损耗值
|
||||||
├── etsnet.py // PSENet中的子网
|
├── etsnet.py // PSENet中的子网
|
||||||
├── fpn.py // PSENet中的子网
|
├── fpn.py // PSENet中的子网
|
||||||
├── resnet50.py // PSENet中的子网
|
├── __init__.py
|
||||||
├── pse // PSENet中的子网
|
├── pse // PSENet中的子网
|
||||||
├── __init__.py
|
|
||||||
├── adaptor.cpp
|
├── adaptor.cpp
|
||||||
├── adaptor.h
|
├── adaptor.h
|
||||||
|
├── __init__.py
|
||||||
├── Makefile
|
├── Makefile
|
||||||
├──config.py // 参数配置
|
├── resnet50 // PSENet中的子网
|
||||||
├──dataset.py // 创建数据集
|
├── __init__.py
|
||||||
├──network_define.py // PSENet架构
|
├── lr_schedule.py // 学习率
|
||||||
├──test.py // 测试脚本
|
├── network_define.py // PSENet架构
|
||||||
├──train.py // 训练脚本
|
├── test.py // 测试脚本
|
||||||
|
├── train.py // 训练脚本
|
||||||
|
|
||||||
```
|
```
|
||||||
|
|
||||||
|
@ -162,7 +165,9 @@ python test.py --ckpt=./device*/ckpt*/ETSNet-*.ckpt
|
||||||
#### 用法
|
#### 用法
|
||||||
|
|
||||||
第一步:单击[此处](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization)下载评估方法。
|
第一步:单击[此处](https://rrc.cvc.uab.es/?ch=4&com=tasks#TextLocalization)下载评估方法。
|
||||||
|
|
||||||
第二步:单击"我的方法"按钮,下载评估脚本。
|
第二步:单击"我的方法"按钮,下载评估脚本。
|
||||||
|
|
||||||
第三步:建议将评估方法根符号链接到$MINDSPORE/model_zoo/psenet/eval_ic15/。如果您的文件夹结构不同,您可能需要更改评估脚本文件中的相应路径。
|
第三步:建议将评估方法根符号链接到$MINDSPORE/model_zoo/psenet/eval_ic15/。如果您的文件夹结构不同,您可能需要更改评估脚本文件中的相应路径。
|
||||||
|
|
||||||
```shell
|
```shell
|
||||||
|
@ -179,35 +184,35 @@ Calculated!{"precision": 0.8147966668299853,"recall":0.8006740491092923,"h
|
||||||
|
|
||||||
### 评估性能
|
### 评估性能
|
||||||
|
|
||||||
| 参数 | PSENet |
|
| 参数 | Ascend |
|
||||||
| -------------------------- | ----------------------------------------------------------- |
|
| -------------------------- | ----------------------------------------------------------- |
|
||||||
| 模型版本 | Inception V1 |
|
| 模型版本 | PSENet |
|
||||||
| 资源 | Ascend 910; CPU 2.60GHz,192内核;内存 755G;系统 Euler2.8 |
|
| 资源 | Ascend 910; CPU 2.60GHz,192内核;内存 755G;系统 Euler2.8 |
|
||||||
| 上传日期 | 2020-09-15 |
|
| 上传日期 | 2020-09-15 |
|
||||||
| MindSpore版本 | 1.0-alpha |
|
| MindSpore版本 | 1.0.0 |
|
||||||
| 数据集 | ICDAR2015 |
|
| 数据集 | ICDAR2015 |
|
||||||
| 训练参数 | start_lr=0.1; lr_scale=0.1 |
|
| 训练参数 | start_lr=0.1; lr_scale=0.1 |
|
||||||
| 优化器 | SGD |
|
| 优化器 | SGD |
|
||||||
| 损失函数 | LossCallBack |
|
| 损失函数 | LossCallBack |
|
||||||
| 输出 | 概率 |
|
| 输出 | 概率 |
|
||||||
| 损失 | 0.35 |
|
| 损失 | 0.35 |
|
||||||
| 速度 | 1卡:444毫秒/步;4卡:446毫秒/步
|
| 速度 | 1卡:444毫秒/步;8卡:446毫秒/步
|
||||||
| 总时间 | 1卡:75.48小时;4卡:18.87小时|
|
| 总时间 | 1卡:75.48小时;8卡:7.11小时|
|
||||||
| 参数(M) | 27.36 |
|
| 参数(M) | 27.36 |
|
||||||
| 微调检查点 | 109.44M (.ckpt file) |
|
| 微调检查点 | 109.44M (.ckpt file) |
|
||||||
| 脚本 | <https://gitee.com/mindspore/mindspore/tree/master/model_zoo/official/cv/psenet> |
|
| 脚本 | <https://gitee.com/mindspore/mindspore/tree/master/model_zoo/official/cv/psenet> |
|
||||||
|
|
||||||
### 推理性能
|
### 推理性能
|
||||||
|
|
||||||
| 参数 | PSENet |
|
| 参数 | Ascend |
|
||||||
| ------------------- | --------------------------- |
|
| ------------------- | --------------------------- |
|
||||||
| 模型版本 | Inception V1 |
|
| 模型版本 | PSENet |
|
||||||
| 资源 | Ascend 910;系统 Euler2.8 |
|
| 资源 | Ascend 910;系统 Euler2.8 |
|
||||||
| 上传日期 | 2020/09/15 |
|
| 上传日期 | 2020/09/15 |
|
||||||
| MindSpore版本 | 1.0-alpha |
|
| MindSpore版本 | 1.0.0 |
|
||||||
| 数据集| ICDAR2015 |
|
| 数据集| ICDAR2015 |
|
||||||
| 输出 | 概率 |
|
| 输出 | 概率 |
|
||||||
| 准确性 | 1卡:81%; 4卡:81% |
|
| 准确性 | 1卡:81%; 8卡:81% |
|
||||||
|
|
||||||
## 使用方法
|
## 使用方法
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue