This project was started on June 2nd, 2006 and is still in its infancy. Nothing is set in stone. Everything from documentation to infrastructure to kernel classes to Component classes needs much more work to be made polished. The names of existing classes are likely to change as the design evolves. You get the picture - it is a time of fluidity for the Conjecture project!
But a baseline implementation does exist, lots of useful infrastructure is already available, and customization at all levels is already supported. It is quite close to being ready for true public consumption, all things considered. And you can help.
If you are seeing this documentation, it is because you may have an interest in contributing to open-source OCR development. Conjecture is designed to make it easy for you to do so at any number of levels. Before Conjecture goes fully public, it just needs a core group of people who can help stablize and extend the framework, brainstorm about desired features, and help get things sufficiently polished.
Most of the high-level documentation is provided via the main web-page. This document hierarchy focuses on per-class design and implementation issues. For an overview of the architecture and design, see the main pages.
See also the high-level documentation.