> I was trying to use the training methods you mention in the Usage > section using the lines data, however the training takes a long time > to complete.
There are a bunch of parameters that you can use to speed up training. The default parameters are set so that you can be fairly certain to get a reasonable result. > Can you tell me what is the average time for it to complete on recent > hardware ? > That is just some benchmark for different types of hardware. I don't remember exactly how long it took when I tested it on that data set; maybe 10 minutes or so on my machine? The default parameters are set for a four core machine. If you have a single core machine, it's going to take four times as long. Training models on a few million characters takes hours. OCRopus 0.5 is going to contain other training methods that scale to larger training sets. For very small training sets, the nearest neighbor classifier may also be reasonable. > Finally, I am very interested about your Decapod project, do you plan > to open-source it as Ocropus?. I really want to be able to try to use > it for deskewing pictures for my project. Yes, it will be open sourced. > Also would like to see Icelandic characters in the default model, they > are only around 10 (áéíóúýþæöð) and in latin1. Our plan is to support all of Latin 1, but we still have to see how to best do that. Training depends on character frequencies, so it tends to be language-specific. Tom --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "ocropus" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/ocropus?hl=en -~----------~----~----~----~------~----~------~--~---
