73459.5 / 29 December 2019

Optic.Ninja Release..BETA

I worked previously on a project to enable me to edit and annotaton datasets. It isn't Currently complete, but it is some what usable. I am excited to release it.

1
2
3
4
It doesn't have any tests.
It currently only supports images.
It's exporting functionality is very limited.
It has no releases yet so you have to build it yourself.

I think it is important that we start to work on an open source local dataset mainipulation tool. When I first started working on this several years ago none existed. I originally created this project mainly to expand on my abilities with C++ and Qt which in that regard I think it is at least partially already a success. Also I used some of the skills I built with this project to start another project to learn more about Qt mobile development, and embedded development. I will be releasing those projects soon as well, but any way let me continue with Optic.Ninja.

I have a roadmap published on the Github page for what I am hoping to accomplish in various time tables. If you would like to use it, suggest new features, provide monetary support, or even hack on it yourself then I welcome it. Contact me via a pull request, or submit an issue on Github.

Optic.Ninja

Optic.Ninja

Screenshot

Optic.Ninja

Description

This project aims to make a simple and efficent dataset editor.

Current Support

1
2
3
4
5
6
- Currently we are targeting image datasets.
- Automatic annotation saving
- CSV Export
- Sqaure annotation marker
- Polyline annotation marker
- Circle annotation marker

Roadmap

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
- Improve "Lazy Loading List View".  Need to add the ability to unload items.
- Improve ListView icon sizing, numbering, and view types.  Also make it resizable.
- Add the ability to copy annotations and copy them between dataset items.
- Add ability to switch to opengl dataset item view.
- Add auto detection after certain number of annotations are added to datset items.
- Add Unit tests
- Add error correction and catching
- Improve loading times and database creation method.
- Add support for multi-part images
- Common dataset importer

Pull requests

Pull requests are welcome. If you have something to contribute please do. Trying to stick to Qt framework, but other libraries are also fine if your implementation is useful.

Build

Linux

Simple install opencv libraris and it should automatically be detected. Tested on Debian/Ubuntu. Then simply build with Qt Creator either from the repos, or download Qt Creator directly from the Qt website.

Windows

Simply clone the repo and build with Qt Creator. You can use MinGW or MSVC.

Mac

I haven't tested building on Mac yet, but it should work in a similar fashion to Linux. I will test this soon.