Cytomine is an open source software developed at the University of Liège, Belgium for collaborative analysis of multi-gigapixel imaging data.
Although the initial focus was on biomedical research with cytology and histology whole-slide images, this is a generic software that might be useful for other types of imaging data from various fields (e.g. botany, geology, astronomy, environmental biology, etc).
Based on modern web, database, and machine learning algorithms, it has been designed to foster active and distributed collaboration between computer vision/machine learning researchers and other scientists dealing with very large images.
Among other functionalities, it implements:
- A web user interface to explore very large images using pyramidal formats
- A web user interfaces to efficiently and collaboratively build large semantic ground-truth datasets from multi-gigapixel images
- A multi-thread implementation of our tree-based learning recognition algorithms (for content-based image retrieval, object classification, image segmentation, and landmark detection
- A web proofreading tools to edit (ours or yours) algorithm predictions
- A RESTful API and software templating mechanisms to easily import/export data with your own software and extend it to your own purposes
- Python and Java clients to easily access the system