In the training process, although I have all the files I need, I have some failure report: Couldn't find a matching blob. Is that normal?
Thanks! ...... APPLY_BOXES: boxfile line 28970/g ((20496,698),(20515,729)): FAILURE! Couldn't f ind a matching blob APPLY_BOXES: Boxes read from boxfile: 29046 Boxes failed resegmentation: 463 ...... APPLY_BOXES: Unlabelled word at :Bounding box=(5908,960)->(5944,971) APPLY_BOXES: Unlabelled word at :Bounding box=(690,962)->(761,994) APPLY_BOXES: Unlabelled word at :Bounding box=(2307,959)->(2345,972) Found 28583 good blobs and 1026 unlabelled blobs in 0 words. 74 remaining unlabelled words deleted. TRAINING ... Font name = arial Generated training data for 5943 words On Tue, Aug 23, 2011 at 7:32 PM, bo gao <[email protected]> wrote: > Hi, All, > > For dictionary: > > I added dictionary for Tessearct 3, but I did not see the output changed. > > Then I try to turn up parameters as told in Wiki page: > > Try upping NON_WERD and GARBAGE_STRING in dict/permute.cpp to maybe 3 or > even 5. > > There is no NON_WERD and GARBAGE_STRING in dict/permute.cpp, should I > refer to segment_penalty_garbage segment_penalty_dict_nonword in > dict/dict.h? > > How can I put more weights on dictionary? > > For training: > > I used the 32 tiff files, but after training the performance degrade, and > the traineddata is much smaller. How should I improve the performance? > Anyone trained a better classifier than provided eng,traineddata? > > Thanks! > -- > > Best, > > Bo > -- Best, Bo -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" 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/tesseract-ocr?hl=en
