i tried this and figured out how to use it sort of. However it constantly crashes and could not get a single satisfactory scan so I only recommend it for masochists. Here are some additional instructions for such:

First of all, don't try the automatic full page recognize. The instructions already state that it does not work. Use the text selection recognize mode only.

Next you have to train it first. Create a jpg doc of some text and use the learn button. Highlight some of the text in small portions and experiment. I used the recognize all check box. in the prefs after clicking "Learn." I am not sure here what triggers the learning process here after selecting the text. In any case I was left with some files in a folder in my "Documents" folder in a subfolder named "DigitEyeOCR." According to the instructions, that is what is important. To do OCR, do NOT choose open first, but click on "Recognize." This will allow you to open jpg files or perhaps other scan files (not PDF's). It then go through an alignment procedure and waits for text selection. After you select the text (recommend small segments), it goes through recognition and then crashes as it creates an output file on your desktop. This output file will have 0 bytes. If you are lucky enough to actually create a file with something in it, then you can open it with the program, edit it, and save it to a more common format. If it crashes, it only crashes a portion of it, so you must quit the main part which is still open, then relaunch to begin again. However if your experience is like mine, I recommend you quit it and drag it to the trash.