CVWS.NET is an open source Computer Vision system capable of finding and solving word searches in images from smartphones.
The overall project is comprised of:
- libCVWS - the library providing the core functionality and is intended for use by developers of other applications
- DemoGUI - a desktop application designed to showcase the project. Allows you to select the processing method for each stage from detecing the word search right through to solving it
- QuantitativeEvaluation - crunches the numbers to determine just how good the overall system is, as well as comparing the performance of different implementations of each stage
- ImageMarkup - database-less storage of data about images containing word searches. Used for training, cross-validation and evaluation of the feature extractors and classifiers that lie at the heart of this project
- DataEntryGUI - a desktop application for entering data about images containing word searches
- UnitTests - self-explanatory. Ensures that the more complicated low-level methods of the project (that could produce deep bugs) are doing what they should be. Tests are re-run before builds to keep things that way!
libCVWS
At the core of the project is libCVWS, a library containing all of the methods necessary to go from an image containing a wordsearch along with a list of word, to a solution. This means that everything needed to build your own word search solving application can be found in this library. libCVWS exposes as much of what’s going on “under the hood” as possible to give developers choice about how each processing stage is handled (e.g. would you rather specify a location for a word search rather than finding it, which method should be used to segment a word search image up into its rows & columns). By exposing this functionality, libCVWS is also of use in other applications that have nothing to do with word searches (e.g. use libCVWS.Imaging to efficiently combine images into a single larger one, or easily set up a Neural Network with common feature extraction techniques by using libCVWS.ClassifierInterfacing).