GithubHelp home page GithubHelp logo

autoseg's Introduction

Introduction to AutoSeg -- an image annotation tool

Attension: the tool doesn't accept path with special characters, such as Chinese characters.

For latest updating information please refer to <ReleaseNotice.txt>

Introduction

Easy and user friendly annotation tool, it supports area annotation, polygon annotation, rectangle annotation.

Annotation of landmarks & points are not available yet.

You may extract imagesTest.7z anywhere in your disk and start testing.

Functions overview are given as follows,

img

In processing of marking, single click to add points, and double click to finish the polygon or spline areas, etc.

After you finished the marking, you may edit the polygon,

  1. double click to add points to the nearest edge
  2. right click to delete extra points

When the polygon is acceptable, click the check button (or press the carriage return button on your keyboard) to accept it and convert it to annotation.

img

img

The annotation files are saves as json format, mainly coco format, you need a category file for all annotations. For Area Annotation, it is binary RLE code, the same as coco json rules. If you need c++ code for your application, please check the mask-api files for detail, please note our mask-api has updated to support both horizontal and vertical scanlines, while coco-api only supports vertical scanlines.

We are now still updating the program, it will support PNG masks, binary masks, and continue to support RLE binary mask(json format); The program will also include more automatical speedy tools for annotation.

Different from coco format (2 level), Autoseg accepts multiple levels of annotation, you may check the menu category->setup to see the results.

Some functions (left row buttons) are planned but not finished yet, luckily these are functions not frequently used.

You can play with functions of superpixels or cannyedge, these functions are a little bit complicated to use, but easy to understand. I just give the demo of using it.

The "Delete" function is given as follows, you make a selection with polygon, polyspline, or manual draw button, when press the del key on your keyboard, any selected area inside will be marked as background.

img

You may read the about-> help if you have comments or requirements about these annotation tool.

Categories should be carefully manipulated, so some enabling/disabling functions are added, as below

[CategoryManipulation.gif]

Automatic Saving

This function is specially design to speed up your annotation.

The annotations will be saved for you when you click next or previous button, so no time will be wasted when you move on to the next image.

Below you can see that the saving box popups only once for the destination folder of the annotations,

[please download the AutoSaving.gif to your computer for a preview].

General short key

A/a -- auto smart mode F/f -- foreground mode B/b -- background mode

carriage return-- accept/transfer current results to annotation

escape - - reset all conditions

Donation

We will keep updating this software, any advice are welcome.

如果您熟悉使用QQ群,可以加入我们的服务群SpaceServices (262375262),我们的工程师会负责解答您的问题。

About us

SpaceSoftwares is a team focused on high accuracy computer vision application. Here we target to provide the highest accuracy in semantic segmentation to auto-driving, medical imaging, general machine learning, tracking, automation, etc.;

We focusing on accuracy improvements, our AI models are widely applied to industrial applications such as precision positioning system, defect detection, classification, etc..

For anything please contact us [email protected]

autoseg's People

Contributors

spaceview avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.