URL: http://www.3igs.org
BIGS is a framework for large scale image processing and machine learning by using computing power in parallel and designed with the following principles:
Simplicity for the experimenter
Simplicity for the algorithm developer
Harness computing power whenever is available, whether on desktops, computer labs, clusters or the cloud
BIGS is designed to exploit coarse grained paralellism through data partitioning and is founded upon two main concepts:
Pipelines, defined by the experimenter, specifying a chain of data processing operations to be performed on input data (images, vectors).
Workers: BIGS agents in charge of resolving the computational load required by pipelines.
Workers can be deployed over desktops, computer labs, clusters, or cloud resources.Workers can be added or removed anytime to an on-going computation. Computing resources need not be committed in advance to any computation, and can join as they become available.