This project aggregates every image it can find based on a keyword search among multiple data sources such as google images, bing images, picasa, flickr, etc.
- Can create datasets for research purposes in matter of minutes
- Easy to add backends
- Automatically prune duplicates (currently being developed)
Please ensure that Python 3 is installed before proceeding. We highly
recommend using a virtual environment (which can be installed using
pip install virtualenv).
git clone https://github.com/soravux/idasc.git
cd idasc
virtualenv venv
source venv/bin/activate
pip install -r requirements.txt
cp config.ini.dist config.ini
nano config.ini
- Bing: http://www.bing.com/dev/en-us/dev-center
- imgur: https://imgur.com/account/settings/apps
- flickr: https://www.flickr.com/services/apps/create/
- Google: https://console.developers.google.com/ (key) and https://www.google.com/cse/ (cx)
python idasc.py [keyword]
python idasc.py --help
Just create a new python module in the backends directory containing a
go(keyword, path) function. This function will receive the keyword entered
by the user and the path where it should download its images. Most backends are
similar and thus could be copied over another similar backend and modified.