Skip to content

DREAMS-lab/deepgis

Repository files navigation

DeepGIS

DeepGIS is an annotation tool for images like cameratraps, UAV imagery, satellite images. It is widely used for annotation, learning and introspection of semantic maps.

DeepGIS is developed using the following technologies

Installation

  • Edit the docker-compose.yml port forwarding to valid ports on your machine.
foo@bar:~$ git clone https://github.com/DREAMS-lab/deepgis.git
foo@bar:~$ cd deepgis
foo@bar:~$ docker-compose up

Configuration

Add a super user in DeepGIS

foo@bar:~$ docker exec -it deepgis_web_1 bash
foo@bar:~$ python manage.py createsuperuser

Add images to DeepGIS

If you have a .tif file, you can use the following command to split the tif into 100 750x750 png images.

foo@bar:~$ python manage.py runscript split_tif cropped_image.tif webclient/static/dataset/ 100

If you are using a map_label app, then add a tiled raster image as the data source. You can add that in the django admin page.

We have provided a few test images in webclient/static/small-tomatoes directory.

foo@bar:~$ python manage.py runscript injectImages /app/webclient/static/small-tomatoes/ tomatoes tomatoes
foo@bar:~$ python manage.py collectstatic

Viewing labels

Labels are stored in the folder specified in the settings file, default being 'labels', which should be in the static-root folder. Labels can also be inspected through the web app using the admin image labels page.

UI Controls

You can resize the browser using standard browser controls (Ctrl+ and Ctrl-). Additionally, the app has sliders to control brightness, contrast, and hue. Adjust these to make the target objects clearer.

Labels can be moved by dragging, and deleted by double-clicking. Choose 'No labels in image' option when there are no labels in the image. Submit when finished labeling an image.

Usage

The webapp is be serving at localhost:8000. The admin page is available at localhost:8000/admin

Visuals

Wildlife annotations in DeepGIS image

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Authors and Acknowledgment

Affiliation: Prof. Jnaneshwar "JD" Das, Distributed Robotic Exploration and Mapping Systems Laboratory, ASU School of Earth and Space Exploration

This project is derived from AgDSS

License

Apache License 2.0

Project Status

Active

About

Annotation, learning, and introspection tool for semantic mapping

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published