oxmap / imglab

Web based tool to label images for objects. So that they can be used to train dlib or other object detectors. Integrated with Face++

Home Page:https://naturalintelligence.github.io/imglab/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Web based tool to label images for object. So that they can be used to train dlib or other object detectors.

Open Source Helpers first-timers-only

Become a Patron! Stubmatic donate button Donate using Liberapay

If you're using the old version of imglab please use this link

Features

ImgLab is platform independent, runs directly from browser, and no prerequisite is needed. It requires very less CPU and memory.

The best thing is that you can use 3rd party libraries to fast annotation process which eventually saves your effort and time both.

Other features are:

  • Designed specially for dlib users. You can easily adjust the order of parts / landmarks/ featurepoints.
  • Opensource and free forever.
  • You can draw feature points, shapes (circle, rectangle, polygon). Other shapes like ellipse, line, curves will be added in future.
  • Small project file so it can be shared over the mail.
  • Import / Export data in multiple formats (Currently dlib XML, and pts format is supported)

How to use

You can either import a file from a URL or from your computer. You can plot the landmark points by yourself or you can request to face++ API to collect the points which gets automatically plotted on the image (You will need to register on face++ to use the API.). If you feel that the result should be improved, you can drag a point to correct location.

ImgLab Tutorial

Version 2

we are working on version 2. We'll be replacing version 1 with version 2 as soon as the primary features are completed. If you want to try it early, check this link : https://naturalintelligence.github.io/imglab/v2.html

Link your work

Using this library if you build some open database which can help others or just to showcase, please raise an issue or PR.

Worth to mention

  • निम्न (NIMN) : Schema aware object compression. 60% more compressed than JSON. 40% more compressed than msgpack.
  • imglab : Web based tool to label images for object. So that they can be used to train dlib or other object detectors. You can integrate 3rd party libraries for fast labeling.
  • fast-lorem-ipsum : Generate lorem ipsum words, sentences, paragraph very quickly.
  • stubmatic : A stub server to mock behaviour of HTTP(s) / REST / SOAP services. You can also mock msgpack, and nimn format in easy way.
  • अनुमार्गक (anumargak) : Amazing fast router for node web servers.
  • fastify-xml-body-parser : Fastify plugin / module to parse XML payload / body into JS object using fast-xml-parser.
  • Grapes : Flexible Regular expression engine (for java) which can be applied on char stream. (under development)

Contributors

About

Web based tool to label images for objects. So that they can be used to train dlib or other object detectors. Integrated with Face++

https://naturalintelligence.github.io/imglab/

License:MIT License


Languages

Language:JavaScript 52.8%Language:HTML 42.8%Language:CSS 4.1%Language:Python 0.4%