The first part is to extract letters and words from an image of a printed page. So the system will take an image of a print text, and output a text of the words in that image. There are so many factors that make this task much easier than recognising the words from a handwritten note.
The first two steps for this task is to build a dataset that is suitable for training, and then train a letter recogniser based on such dataset. For the training part, you are free to use any of the methods that we have already used in the lab (svm, knn, neural network etc).
The second task is to create a similar engine, but this time the engine should work on handwritten notes. Again, your task is to create the training dataset, using your own handwritten notes, and then train a letter recog- niser on these data. Experiment with different classifiers, and report your findings, comparing with the recogniser that works on printed pages.
5 freelancers are bidding on average £371 for this job
I recently received my masters degree in Computational Neuroscience (mostly vision studies). I also just completed a project very similar to this one. I would be happy to do this for you. Thanks, Joe