LabelingSystem/web/libs/editor/examples/image_magic_wand/START.md

Magic Wand for Image Segmentation

Install

Linux & Ubuntu guide

Install python and virtualenv

# install python and virtualenv 
apt install python3.6
pip3 install virtualenv

# setup python virtual environment 
virtualenv -p python3 env3
source env3/bin/activate

# install requirements 
cd backend
pip install -r requirements.txt

Cross Domain Image Access

Note that if you are storing images that you'd like to apply the Magic Wand to cross-domain, such as on Google Storage Buckets, you will have to enable CORS headers for the storage buckets to enable cross-domain pixel access so that the Magic Wand can get the raw pixel data to threshold. By default browsers block JavaScript from accessing pixel-level image data unless the right CORS headers are set.

As an example, if you wanted to configure a Google Storage Bucket with the right headers, you might do the following:

gsutil cors set gcp_cors_config.json gs://BUCKET-NAME

Note that in the gcp_cors_config.json example given in this directory that we have set origin to *, which means all origins can access that data, as well as set responseHeader to *, which means all HTTP response headers can be accessed. In a real scenario you probably want to think through the security ramifications of this for your own particular Label Studio setup.

Start

Magic Wand for image segmentation:

python server.py -c config.json -l ../examples/image_magic_wand/config.xml -i ../examples/image_magic_wand/tasks.json -o output