Skip to content

My own version to annotate dataset for YOLO format (Including multi-class labeling on the same image)

Notifications You must be signed in to change notification settings

ivder/Yolo_MultiClass_LabelTool

Repository files navigation

My own version of labeling tool for YOLO format (support Multi Class labeling on the same image)

Main Program forked from puzzledqs/BBox-Label-Tool
Converter to Yolo format forked from [ManivannanMurugavel/YOLO-Annotation-Tool] (https://github.com/ManivannanMurugavel/YOLO-Annotation-Tool)

Feature

  1. Multi-class support
  2. Support .jpg and .png format
  3. Built in YOLO format converter

Additional Feature

  1. Skip button to skip labeling on unwanted image
  2. Add Save and Load Checkpoint
  3. Remove class confirm button (set value directly from combobox)
  4. Add Convert to YOLO format button (no need to run external program)
  5. Load image using directory browser instead of user input

Data Organization

LabelTool
|
|--main.py # source code for the tool
|--Images/ # direcotry containing the images to be labeled
   |--Sample/ # project/directory name
|--Result/ # direcotry for the labeling results
   |--Sample/ # result txt according to project name
|--Result_YOLO/ # converted to YOLO format

Yolo Annotator

A simple tool for labeling object bounding boxes in images, implemented with Python Tkinter.

Dependency

python 2.7 win 32bit PIL-1.1.7.win32-py2.7

Usage

  1. For multi-class task, modify 'class.txt' with your own class-candidates and before labeling bbox, choose the 'Current Class' in the Combobox or by pressing 1-9 on your keyboard.
  2. run python main.py
  3. click LoadImage, select a folder that contains list of images.
  4. To create a new bounding box, left-click to select the first vertex. Moving the mouse to draw a rectangle, and left-click again to select the second vertex.
  • To cancel the bounding box while drawing, just press Esc or s.
  • To delete a existing bounding box, select it from the listbox, and click Clear or r.
  • To delete all existing bounding boxes in the image, simply click ClearAll.
  1. After finishing one image, click Next or d to advance. Likewise, click Prev or a to reverse. Or, input the index and click Go to navigate to an arbitrary image.
    • The labeling result will be saved in Labels/[folder name]/.. if and only if the 'Next' button is clicked.
    • Checkpoint of last Image Number will be saved when 'Next' button is clicked.
  2. Click Skip if you want to skip unwanted image from directory and skip the annotation for that image (skipped image path will be saved in log/skip.txt)
  3. Click ConvertYOLO button or to convert the labeling result to YOLO format. The result will be saved in Result_YOLO/[folder name]/..

Output

Example

Result (bbox coodrdinates):

2
99 17 571 436 dog
733 60 988 320 cat

Result_YOLO (yolo format) :

0 0.279166666667 0.359523809524 0.393333333333 0.665079365079
1 0.717083333333 0.301587301587 0.2125 0.412698412698

About

My own version to annotate dataset for YOLO format (Including multi-class labeling on the same image)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages