Build and link against Tesseract for OpenCV text module

[xtrntr]

Help on module cv2.text in cv2:

NAME
cv2.text

FUNCTIONS
OCRBeamSearchDecoder_create(...)
OCRBeamSearchDecoder_create(classifier, vocabulary, transition_probabilities_table, emission_probabilities_table[, mode[, beam_size]]) -> retval

OCRHMMDecoder_create(...)
    OCRHMMDecoder_create(classifier, vocabulary, transition_probabilities_table, emission_probabilities_table[, mode]) -> retval

OCRTesseract_create(...)
    OCRTesseract_create([, datapath[, language[, char_whitelist[, oem[, psmode]]]]]) -> retval

createOCRHMMTransitionsTable(...)
    createOCRHMMTransitionsTable(vocabulary, lexicon) -> retval

loadOCRBeamSearchClassifierCNN(...)
    loadOCRBeamSearchClassifierCNN(filename) -> retval

loadOCRHMMClassifierCNN(...)
    loadOCRHMMClassifierCNN(filename) -> retval

loadOCRHMMClassifierNM(...)
    loadOCRHMMClassifierNM(filename) -> retval

DATA
ERFILTER_NM_IHSGRAD = 1
ERFILTER_NM_IHSGrad = 1
ERFILTER_NM_RGBLGRAD = 0
ERFILTER_NM_RGBLGrad = 0
ERGROUPING_ORIENTATION_ANY = 1
ERGROUPING_ORIENTATION_HORIZ = 0
OCR_DECODER_VITERBI = 0
OCR_LEVEL_TEXTLINE = 1
OCR_LEVEL_WORD = 0

FILE
(built-in)

[patricksnape]

Sorry, I have no idea what your question is, please be more specific.

[xtrntr]

opencv/opencv_contrib#798

[patricksnape]

So this is a request to build opencv against tesseract? At the moment I have no bandwidth to do that but if you can make a conda package for teserract I would be happy to link against it!

[xtrntr]

Could you walk me through the steps to do that? I'm interested at taking a stab.

[patricksnape]

Take a look at the conda-build documentation and the meta.yaml here. You can probably base the recipe off of the homebrew package as a good starting place.