This project will eventually develop a complete Bengali language module for ocropus, including segmentation, recognition, language modeling, dictionary access, and grounded truth sub-modules. At the moment it's basically some of the existing ocropus code modified to work with Bengali and a very limited grounded truth library for neural network training and overall recognition evaluation. To compile and run this project currently requires an existing ocropus source code tree and all of the system libraries needed to get ocropus to compile correctly.