From 980b4da65ada170a91057422772b8b2b71618f7a Mon Sep 17 00:00:00 2001 From: lart Date: Sat, 16 Dec 2023 21:09:43 +0800 Subject: [PATCH] Fixed and simplified the description of the readme files. --- readme.md | 29 ++++++----------------------- readme_zh.md | 27 +++++++-------------------- 2 files changed, 13 insertions(+), 43 deletions(-) diff --git a/readme.md b/readme.md index 6eda6f5..0dc1344 100644 --- a/readme.md +++ b/readme.md @@ -12,7 +12,7 @@ ## Introduction -A simple and efficient implementation of SOD metrics +A simple and efficient implementation of SOD metrics. - Based on `numpy` and `scipy` - Verification based on Fan's matlab code @@ -23,7 +23,7 @@ Your improvements and suggestions are welcome. ### Related Projects -- A Python-based salient object detection and video object segmentation evaluation toolbox. +- [PySODEvalToolkit](https://github.com/lartpang/PySODEvalToolkit): A Python-based Evaluation Toolbox for Salient Object Detection and Camouflaged Object Detection ### Supported Metrics @@ -35,7 +35,7 @@ Your improvements and suggestions are welcome. | Multi-Scale IoU | bin | | `MSIoU` | | E-measure ($E_{m}$) | max,avg,adp | | `Emeasure` | | F-measure (old) ($F_{beta}$) | max,avg,adp | | `Fmeasure` | -| F-measure (old) ($F_{beta}$, $F_{1}$) | max,avg,adp,bin | bin | `FmeasureV2`+`FmeasureHandler` | +| F-measure (new) ($F_{beta}$, $F_{1}$) | max,avg,adp,bin | bin | `FmeasureV2`+`FmeasureHandler` | | BER | max,avg,adp,bin | bin | `FmeasureV2`+`BERHandler` | | Dice | max,avg,adp,bin | bin | `FmeasureV2`+`DICEHandler` | | FPR | max,avg,adp,bin | bin | `FmeasureV2`+`FPRHandler` | @@ -48,24 +48,6 @@ Your improvements and suggestions are welcome. | Specificity | max,avg,adp,bin | bin | `FmeasureV2`+`SpecificityHandler` | | TNR | max,avg,adp,bin | bin | `FmeasureV2`+`TNRHandler` | | TPR | max,avg,adp,bin | bin | `FmeasureV2`+`TPRHandler` | -## TODO List - -- [x] Speed up the calculation of Emeasure. -- [x] Add the necessary documentation for different functions. - -## Compared with [Matlab Code from Fan](https://github.com/DengPingFan/CODToolbox) - -In our comparison (the test code can be seen under the `test` folder), the result is consistent with Fan's code, as follows: - -```text -ours: Smeasure:0.903; wFmeasure:0.558; MAE:0.037; adpEm:0.941; meanEm:0.957; maxEm:0.967; adpFm:0.582; meanFm:0.577; maxFm:0.589 -matlab: Smeasure:0.903; wFmeasure:0.558; MAE:0.037; adpEm:0.941; meanEm:0.957; maxEm:0.967; adpFm:0.582; meanFm:0.577; maxFm:0.589. -``` - -**NOTE** - -- The matlab code needs to change `Bi_sal(sal>threshold)=1;` to `Bi_sal(sal>=threshold)=1;` in . For related discussion, please see [the issue](https://github.com/DengPingFan/CODToolbox/issues/1). -- 2021-12-20 (version `1.3.0`): Due to the difference between numpy and matlab, in version `1.2.x`, there are very slight differences on some metrics between the results of the matlab code and ours. The [recent PR](https://github.com/lartpang/PySODMetrics/pull/3) alleviated this problem. However, there are still very small differences on E-measure. The results in most papers are rounded off to three or four significant figures, so, there is no obvious difference between the new version and the version `1.2.x` for them. ## Usage @@ -81,8 +63,9 @@ The core files are in the folder `py_sod_metrics`. ## Reference -- - - By DengPingFan() +- [Matlab Code](https://github.com/DengPingFan/CODToolbox) by DengPingFan(): In our comparison (the test code can be seen under the `test` folder), the result is consistent with the code. + - The matlab code needs to change `Bi_sal(sal>threshold)=1;` to `Bi_sal(sal>=threshold)=1;` in . For related discussion, please see [the issue](https://github.com/DengPingFan/CODToolbox/issues/1). + - 2021-12-20 (version `1.3.0`): Due to the difference between numpy and matlab, in version `1.2.x`, there are very slight differences on some metrics between the results of the matlab code and ours. The [recent PR](https://github.com/lartpang/PySODMetrics/pull/3) alleviated this problem. However, there are still very small differences on E-measure. The results in most papers are rounded off to three or four significant figures, so, there is no obvious difference between the new version and the version `1.2.x` for them. - ```text diff --git a/readme_zh.md b/readme_zh.md index 960b0fd..f2b88e0 100644 --- a/readme_zh.md +++ b/readme_zh.md @@ -12,10 +12,10 @@ ## 介绍 -一份简单有效的 SOD 指标实现 +一份简单有效的 SOD 指标实现。 - 基于`numpy`和极少量`scipy.ndimage`代码 -- 基于 DengPing Fan +- 基于 DengPing Fan 进行对比验证 - 结构简单，易于扩展 - 代码轻量且快速 @@ -23,7 +23,7 @@ ### 相关项目 -- A Python-based salient object detection and video object segmentation evaluation toolbox. +- [PySODEvalToolkit](https://github.com/lartpang/PySODEvalToolkit): A Python-based Evaluation Toolbox for Salient Object Detection and Camouflaged Object Detection ### 支持的指标 @@ -35,7 +35,7 @@ | Multi-Scale IoU | bin | | `MSIoU` | | E-measure ($E_{m}$) | max,avg,adp | | `Emeasure` | | F-measure (old) ($F_{beta}$) | max,avg,adp | | `Fmeasure` | -| F-measure (old) ($F_{beta}$, $F_{1}$) | max,avg,adp,bin | bin | `FmeasureV2`+`FmeasureHandler` | +| F-measure (new) ($F_{beta}$, $F_{1}$) | max,avg,adp,bin | bin | `FmeasureV2`+`FmeasureHandler` | | BER | max,avg,adp,bin | bin | `FmeasureV2`+`BERHandler` | | Dice | max,avg,adp,bin | bin | `FmeasureV2`+`DICEHandler` | | FPR | max,avg,adp,bin | bin | `FmeasureV2`+`FPRHandler` | @@ -49,20 +49,6 @@ | TNR | max,avg,adp,bin | bin | `FmeasureV2`+`TNRHandler` | | TPR | max,avg,adp,bin | bin | `FmeasureV2`+`TPRHandler` | -## 与[范的 Matlab 代码](https://github.com/DengPingFan/CODToolbox)的比较 - -在我们的测试中 (测试代码可见`test`文件夹下内容)，结果与 Fan 的代码一致，如下: - -```text -ours: Smeasure:0.903; wFmeasure:0.558; MAE:0.037; adpEm:0.941; meanEm:0.957; maxEm:0.967; adpFm:0.582; meanFm:0.577; maxFm:0.589 -matlab: Smeasure:0.903; wFmeasure:0.558; MAE:0.037; adpEm:0.941; meanEm:0.957; maxEm:0.967; adpFm:0.582; meanFm:0.577; maxFm:0.589. -``` - -**注意** - -- matlab 代码需要将 的`Bi_sal(sal>threshold)=1;`改为` Bi_sal(sal>=threshold)=1;`。细节可见 [相关讨论](https://github.com/DengPingFan/CODToolbox/issues/1)。 -- 2021-12-20 (Version `1.3.0`)：由于 numpy 和 matlab 的不同，在 `1.2.x` 版本中，matlab 代码的结果与我们的结果在某些指标上存在非常细微的差异。[最近的 PR](https://github.com/lartpang/PySODMetrics/pull/3) 缓解了这个问题。但是，在 E-measure 上仍然存在非常小的差异。大多数论文中的结果都四舍五入到三四位有效数字，因此，新版本与“1.2.x”版本之间没有明显差异。 - ## 使用 核心文件在文件夹 `py_sod_metrics` 中。 @@ -77,8 +63,9 @@ matlab: Smeasure:0.903; wFmeasure:0.558; MAE:0.037; adpEm:0.941; meanEm:0.957; m ## 参考 -- - - By DengPingFan() +- [Matlab Code](https://github.com/DengPingFan/CODToolbox) by DengPingFan(): 在我们的测试中 (测试代码可见`test`文件夹下内容)，结果与 Fan 的代码一致。 + - matlab 代码需要将 的`Bi_sal(sal>threshold)=1;`改为` Bi_sal(sal>=threshold)=1;`。细节可见 [相关讨论](https://github.com/DengPingFan/CODToolbox/issues/1)。 + - 2021-12-20 (Version `1.3.0`)：由于 numpy 和 matlab 的不同，在 `1.2.x` 版本中，matlab 代码的结果与我们的结果在某些指标上存在非常细微的差异。[最近的 PR](https://github.com/lartpang/PySODMetrics/pull/3) 缓解了这个问题。但是，在 E-measure 上仍然存在非常小的差异。大多数论文中的结果都四舍五入到三四位有效数字，因此，新版本与“1.2.x”版本之间没有明显差异。 - ```text