ImageAnnotater is an annotation tool to label images with bounding boxes, points, or lines in a quick way.
- python 3
- PyGObject (pip3 install pygobject)
- Open a terminal in the folder where OdmConverter shall be.
- Clone the repository:
git clone git@github.com:egemose/ImageAnnotater.git
- Install homebrew (https://brew.sh/)
- Install python3 with homebrew
brew install python3 - Install PyGObject with homebrew
brew install pygobject3 --with-python3 gtk+3 - Install gnome-icon-theme with homebrew
brew install gnome-icon-theme
- Open a terminal in the folder where OdmConverter shall be.
- Clone the repository:
git clone git@github.com:egemose/ImageAnnotater.git
python annotateImages.py [-h] [-i str] [-t str] [-p str]
GUI to annotate images.
optional arguments:
-h, --help show this help message and exit
-i str, --images str Folder with images (str).
-t str, --types str File with point types in csv (str).
-p str, --points str File of saved points in csv (str).
Open a image directory with the first button in the toolbox (ctrl-O) or only a single image with the next button. (ctrl-I)
To open the previous image in the folder use 3. button (ctrl-shift-P) To open the next image in the folder use 4. button (ctrl-shift-N).
Open a csv file with the different types of points on the 5. button (ctrl-T). Choose the desired type of point in the drop-down (9. button) or using the numbers.
Open a csv file with markings on the 7. button (ctrl-M). save a csv file with the markings on the 6. button (ctrl-S). save as can be achieved by (ctrl-shift-S)
Use the up and down arrows button (8. button) to switch between the original image and a computer segmented image. (ctrl-<)
Use the 10. button to switch between making lines and bounding boxes. (ctrl-B)
left click to make a marking or left drag to make a size marking. right click to remove a marking or size marking (with the mouse over one of the circles). ctrl left click also removes markings. Middle click and drag to scroll.
zoom in (ctrl-+) and zoom out (ctrl--) on the + or - buttons or zoom to 100% using the 1 button (ctrl-0). The zoom level can also be adjusted on the slider, by dragging or scrolling the mouse wheel when over the slider. Page Up, Page Down, Home, End and ctrl-arrow-key can also be used to zoom. (can happen that the slider need to be click beforehand) Zooming can take some time and the progress bar to the left will show the progress and display "done" when it is finished.
To the left is a table holding a summary of the different images and how many points that have been added and of what type. It also shows the last marking made.
In the top left is a status-bar showing relevant information.
In the app menu bar all actions can be found and there corresponding shortcut.
header: color, type
example:
color, type
FF0000, Scotch broom
0000FF, Giant hogweed
header: image,type,x,y,size,angle,red,green,blue,alpha
example:
image, type, x1, y1, x2, y2, box, red, green, blue, alpha
./DJI_0001.JPG,Scotch broom,100,200,,,False,0.0,0.0,1.0,1
./DJI_0001.JPG,Scotch broom,100,200,200,400,False,0.0,0.0,1.0,1
Written by Henrik Dyrberg Egemose (hesc@mmmi.sdu.dk) as part of the InvaDrone project a research project by the University of Southern Denmark UAS Center (SDU UAS Center).
This project is licensed under the 3-Clause BSD License - see the LICENSE file for details.