Linux CPU |
Linux GPU |
Mac OS CPU |
Windows CPU |
Windows GPU |
|---|---|---|---|---|
Update 23-10-2017: It's an image algorithm toolbox entirely written by myself.
MIT license. Contributions welcome.
Note :
Each sub-tool independent of each other, can be used alone.
各子工具之间互相独立，可单独使用。
Image_Algorithm_Toolbox/ root dir 根目录
|
|
+-- Distributing_Tool/ image distributing tool 图片数据分发工具
| |
| +-- distributing.py main program 主程序
| |
| +-- src/ pics to be distributed (need build by yourself) 存放待分发的图片集(需自己新建)
| |
| +-- dst/ pics be distributed (automatically build by program) 存放已分发的图片集(程序自动新建)
|
|
+-- Enhancing_Tool/ image enhancing tool 图片数据增强工具
| |
| +-- sync_enhancing.py main program 主程序
| |
| +-- src/ pics to be enhanced (need build by yourself) 存放待增强的图片集(需自己新建)
| |
| +-- dst/ pics be enhanced (automatically build by program) 存放已增强的图片集(程序自动新建)
|
|
+-- Formatting_Tool/ format conversion tool 格式转换工具
| |
| +-- format_conversion.py main program 主程序
| |
| +-- src/ pics to be formatted (need build by yourself) 存放待格式转换的文件(需自己新建)
| |
| +-- dst/ pics be formatted (automatically build by program) 存放同名的已转换文件(程序自动新建)
|
|
+-- Labeling_Tool/ labeling tool 图片标记工具
| |
| +-- labeling.py main program 主程序
| |
| +-- src/ pics to be marked (need build by yourself) 存放待标注的图片(需自己新建)
| |
| +-- dst/ pics be marked (automatically build by program) 存放同名的已标注文件(程序自动新建)
|
|
+-- Resizing_Tool/ resizing tool 图片缩放工具
| |
| +-- pic_resizing.py main program 1 主程序1(只缩放图片)
| |
| +-- sync_resizing.py main program 2 主程序2(同步缩放图片和标记文件)
| |
| +-- src/ pics to be resized (need build by yourself) 存放待缩放的数据集(需自己新建)
| |
| +-- dst/ pics be resized (automatically build by program) 存放同名的已缩放数据集(程序自动新建)
|
|
+-- README.md manual of project 说明手册
|
|
+-- LICENSE.md license of project 许可证
|
|
+-- requirements.txt environment required for this program 环境要求
cd Distributing_Tool/；- Build
src/folder and copy dataset into it； - Run
distributing.py； - Processed images automatically be saved into
dst/folder.
cd Enhancing_Tool/；- Build
src/folder and copy dataset into it； - Run
sync_enhancing.py； - Processed images automatically be saved into
dst/folder.
cd Formatting_Tool/；- Build
src/folder and copy dataset into it； - Run
format_conversion.py； - Processed images automatically be saved into
dst/folder.
cd Labeling_Tool/；- Build
src/folder and copy dataset into it； - Run
labeling.py； - Processed images automatically be saved into
dst/folder.
cd Resizing_Tool/；- Build
src/folder and copy dataset into it； - Run
pic_resizing.pyorsync_resizing.py； - Processed images automatically be saved into
dst/folder.
Open the labeling tool:
Input the name of sub folder:
Automatically load corresponding Image:
Automatically create document of the same name:
Specify the coordinates use single left-click, every two clicks to determine a diagonal box. Right click cancel.:
Automatically create messages of marked boxes:
Please enjoy the labeling process (≧▽≦)y :
- Python3.x
- OpenCV3.x
In labeling tool, AttributeError may met when using python2.
- Replace the head lines:
import tkinter as tk
import tkinter.messageboxwith:
import Tkinter as tk
from Tkinter import *
import tkMessageBox- Replace
tk.messagebox.askyesno(appear twice in line 241、255) withtkMessageBox.askyesno.
Then error will be fixed.
In labeling tool, since the window image I designed is auto-scaled, the original mark's coordinate position may shift slightly if you repeatedly access the image. But for high-definition pictures, this tiny offset is often negligible.