https://github.com/jeffffffli/labelkeypoint
Keypoints label tools
https://github.com/jeffffffli/labelkeypoint
Last synced: 2 months ago
JSON representation
Keypoints label tools
- Host: GitHub
- URL: https://github.com/jeffffffli/labelkeypoint
- Owner: jeffffffli
- License: gpl-3.0
- Created: 2018-02-28T09:52:24.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2018-07-17T04:32:16.000Z (almost 7 years ago)
- Last Synced: 2025-03-28T01:14:23.349Z (3 months ago)
- Language: Python
- Size: 6.4 MB
- Stars: 89
- Watchers: 3
- Forks: 26
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
labelme: Image Annotation Tool with Python
==========================================
[](https://pypi.python.org/pypi/labelme)
[](https://travis-ci.org/wkentaro/labelme)
[](https://hub.docker.com/r/wkentaro/labelme)
Requirements
------------
- Ubuntu / macOS / Windows
- Python2 / Python3
- [PyQt4 / PyQt5](http://www.riverbankcomputing.co.uk/software/pyqt/intro)
Installation
------------
**Anaconda**
You need install [Anaconda](https://www.continuum.io/downloads), then run below:
For linux and Mac
```bash
# python2
cd $root_dir_of_labelme
conda create --name=labelme python=2.7
source activate labelme
conda install pyqt
pip install labelme
```
For windows
```bash
# python2
cd $root_dir_of_labelme
conda create --name=labelme python=2.7
activate labelme
conda install pyqt
pip install labelme
```
Usage
-----
**准备**
下载[`coco_anno`](https://drive.google.com/open?id=1_1_E9dr-X33HpmU5Ffo60NCOyrS-ATSB)文件夹，把文件夹路径复制到`labelKeypoint/labelme/app.py`的第100行。比如，我的路径是`/mnt/c/Users/ljf_l/Desktop/coco_anno`，则修改如下
``` python
self.datadir = '/mnt/c/Users/ljf_l/Desktop/coco_anno'
```
**Install**
```bash
# Install
$ pip uninstall labelme
$ python setup.py install
```
**运行**
```bash
# Run
$ labelme # Open GUI
```
**标注**
打开软件之后，选择右侧的open，开始标记
标注的主界面如下
左上方第一个数字表示当前标注的图片序号，`person：`表示当前标记的人的序号，`progress：`表示当前图片已经标记好的人的数量。
点击鼠标左键即可开始标记。
选择`Head`表示标注头部，`Neck`标注脖子。对于没办法预测的部位，可以不标注。如果头和脖子都没有，请在框内任意点击一个位置，选择`Empty`，表示当前人的头和脖子均难以预测，无法标记。
如果对一个人的相同部位重复标记，会弹出提示窗口，关闭即可。
选择`next person`或`prev person`可以切换到不同的人进行标记。
如果标记错误，可选择`Edit`后，在右侧选择相应的部位，进行删除
然后点击`Create`重新进入标注状态
当所有人都标记完后，选择`next`标注下一张图。此时程序会自动保存已经标记好的图片。
中途如果程序关闭，进度会自动保存，下次点击`open`会从直接上次未标注完成的图片出打开，继续标注。
**标注例子**
Acknowledgement
---------------
This repo is the fork of [mpitid/pylabelme](https://github.com/mpitid/pylabelme),
whose development has already stopped.