https://github.com/MrZander/YoloMarkNet

Darknet YOLOv2/3 annotation tool written in C#/WPF
https://github.com/MrZander/YoloMarkNet

annotation-tool yolo-v3 yolov2 yolov3

Last synced: about 1 month ago
JSON representation

Darknet YOLOv2/3 annotation tool written in C#/WPF

• Host: GitHub
• URL: https://github.com/MrZander/YoloMarkNet
• Owner: MrZander
• License: unlicense
• Created: 2018-04-24T01:29:59.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-05-06T22:29:22.000Z (over 6 years ago)
• Last Synced: 2024-08-02T01:25:11.350Z (4 months ago)
• Topics: annotation-tool, yolo-v3, yolov2, yolov3
• Language: C#
• Size: 37.1 KB
• Stars: 19
• Watchers: 2
• Forks: 9
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-object-detection-datasets - MrZander/YoloMarkNet
• awesome-yolo-object-detection - MrZander/YoloMarkNet
• awesome-yolo-object-detection - MrZander/YoloMarkNet

README

        
# YoloMarkNet

WIP, inspired by https://github.com/AlexeyAB/Yolo_mark

I wanted to be able to delete regions, so I made this.

Current features:

* Annotate multiple classes at once

* Outputs to a format that will work for training darknet

* Delete bounding boxes without having to start from scratch

* Drag and Resize bounding boxes

![Example](https://thumbs.gfycat.com/SecretInfantileGrassspider-size_restricted.gif)

1.) Build the project

2.) Place images into Debug/data/img or Release/data/img depending on what config you built under

3.) Put your class names into obj.names and update obj.data to have the correct number of classes

4.) Run YoloMarkNet.exe

5.) Select your class from the panel on the left

6.) Highlight regions in the images

You can view AlexeyAB's repos for instructions on how to train with the generated data

It needs some work still, I plan to add more features:

* ~~Handling scaling better.  (Entire canvas is scaled right now, doesn't work well with large or tiny images)~~

* Select regions point-by-point and find a best fitting rect

* Other output formats

* ~~Adjustable rect sizes (drag edges, move the rectangles around, etc)~~