From 7b8c476700f995c81a43ec46a14efdb42ea11b32 Mon Sep 17 00:00:00 2001 From: AlisaLinUwU Date: Sun, 26 Jan 2025 11:44:59 +0500 Subject: Initialize --- src/main/resources/tessdata/configs/alto | 1 + src/main/resources/tessdata/configs/ambigs.train | 7 +++++++ src/main/resources/tessdata/configs/api_config | 1 + src/main/resources/tessdata/configs/bigram | 5 +++++ src/main/resources/tessdata/configs/box.train | 12 ++++++++++++ src/main/resources/tessdata/configs/box.train.stderr | 13 +++++++++++++ src/main/resources/tessdata/configs/digits | 1 + src/main/resources/tessdata/configs/get.images | 1 + src/main/resources/tessdata/configs/hocr | 2 ++ src/main/resources/tessdata/configs/inter | 2 ++ src/main/resources/tessdata/configs/kannada | 4 ++++ src/main/resources/tessdata/configs/linebox | 2 ++ src/main/resources/tessdata/configs/logfile | 1 + src/main/resources/tessdata/configs/lstm.train | 11 +++++++++++ src/main/resources/tessdata/configs/lstmbox | 1 + src/main/resources/tessdata/configs/lstmdebug | 4 ++++ src/main/resources/tessdata/configs/makebox | 1 + src/main/resources/tessdata/configs/page | 3 +++ src/main/resources/tessdata/configs/pdf | 1 + src/main/resources/tessdata/configs/quiet | 1 + src/main/resources/tessdata/configs/rebox | 2 ++ src/main/resources/tessdata/configs/strokewidth | 12 ++++++++++++ src/main/resources/tessdata/configs/tsv | 1 + src/main/resources/tessdata/configs/txt | 3 +++ src/main/resources/tessdata/configs/unlv | 2 ++ src/main/resources/tessdata/configs/wordstrbox | 1 + 26 files changed, 95 insertions(+) create mode 100644 src/main/resources/tessdata/configs/alto create mode 100644 src/main/resources/tessdata/configs/ambigs.train create mode 100644 src/main/resources/tessdata/configs/api_config create mode 100644 src/main/resources/tessdata/configs/bigram create mode 100644 src/main/resources/tessdata/configs/box.train create mode 100644 src/main/resources/tessdata/configs/box.train.stderr create mode 100644 src/main/resources/tessdata/configs/digits create mode 100644 src/main/resources/tessdata/configs/get.images create mode 100644 src/main/resources/tessdata/configs/hocr create mode 100644 src/main/resources/tessdata/configs/inter create mode 100644 src/main/resources/tessdata/configs/kannada create mode 100644 src/main/resources/tessdata/configs/linebox create mode 100644 src/main/resources/tessdata/configs/logfile create mode 100644 src/main/resources/tessdata/configs/lstm.train create mode 100644 src/main/resources/tessdata/configs/lstmbox create mode 100644 src/main/resources/tessdata/configs/lstmdebug create mode 100644 src/main/resources/tessdata/configs/makebox create mode 100644 src/main/resources/tessdata/configs/page create mode 100644 src/main/resources/tessdata/configs/pdf create mode 100644 src/main/resources/tessdata/configs/quiet create mode 100644 src/main/resources/tessdata/configs/rebox create mode 100644 src/main/resources/tessdata/configs/strokewidth create mode 100644 src/main/resources/tessdata/configs/tsv create mode 100644 src/main/resources/tessdata/configs/txt create mode 100644 src/main/resources/tessdata/configs/unlv create mode 100644 src/main/resources/tessdata/configs/wordstrbox (limited to 'src/main/resources/tessdata/configs') diff --git a/src/main/resources/tessdata/configs/alto b/src/main/resources/tessdata/configs/alto new file mode 100644 index 0000000..0dd12a7 --- /dev/null +++ b/src/main/resources/tessdata/configs/alto @@ -0,0 +1 @@ +tessedit_create_alto 1 diff --git a/src/main/resources/tessdata/configs/ambigs.train b/src/main/resources/tessdata/configs/ambigs.train new file mode 100644 index 0000000..23035a1 --- /dev/null +++ b/src/main/resources/tessdata/configs/ambigs.train @@ -0,0 +1,7 @@ +tessedit_ambigs_training 1 +load_freq_dawg 0 +load_punc_dawg 0 +load_system_dawg 0 +load_number_dawg 0 +ambigs_debug_level 3 +load_fixed_length_dawgs 0 diff --git a/src/main/resources/tessdata/configs/api_config b/src/main/resources/tessdata/configs/api_config new file mode 100644 index 0000000..5cd6ec0 --- /dev/null +++ b/src/main/resources/tessdata/configs/api_config @@ -0,0 +1 @@ +tessedit_zero_rejection T diff --git a/src/main/resources/tessdata/configs/bigram b/src/main/resources/tessdata/configs/bigram new file mode 100644 index 0000000..5d6c2d0 --- /dev/null +++ b/src/main/resources/tessdata/configs/bigram @@ -0,0 +1,5 @@ +load_bigram_dawg True +tessedit_enable_bigram_correction True +tessedit_bigram_debug 3 +save_raw_choices True +save_alt_choices True diff --git a/src/main/resources/tessdata/configs/box.train b/src/main/resources/tessdata/configs/box.train new file mode 100644 index 0000000..d39f268 --- /dev/null +++ b/src/main/resources/tessdata/configs/box.train @@ -0,0 +1,12 @@ +disable_character_fragments T +file_type .bl +textord_fast_pitch_test T +tessedit_zero_rejection T +tessedit_minimal_rejection F +tessedit_write_rep_codes F +edges_children_fix F +edges_childarea 0.65 +edges_boxarea 0.9 +tessedit_resegment_from_boxes T +tessedit_train_from_boxes T +textord_no_rejects T diff --git a/src/main/resources/tessdata/configs/box.train.stderr b/src/main/resources/tessdata/configs/box.train.stderr new file mode 100644 index 0000000..82754e9 --- /dev/null +++ b/src/main/resources/tessdata/configs/box.train.stderr @@ -0,0 +1,13 @@ +file_type .bl +#tessedit_use_nn F +textord_fast_pitch_test T +tessedit_zero_rejection T +tessedit_minimal_rejection F +tessedit_write_rep_codes F +edges_children_fix F +edges_childarea 0.65 +edges_boxarea 0.9 +tessedit_resegment_from_boxes T +tessedit_train_from_boxes T +#textord_repeat_extraction F +textord_no_rejects T diff --git a/src/main/resources/tessdata/configs/digits b/src/main/resources/tessdata/configs/digits new file mode 100644 index 0000000..6a329f8 --- /dev/null +++ b/src/main/resources/tessdata/configs/digits @@ -0,0 +1 @@ +tessedit_char_whitelist 0123456789-. diff --git a/src/main/resources/tessdata/configs/get.images b/src/main/resources/tessdata/configs/get.images new file mode 100644 index 0000000..7d00b61 --- /dev/null +++ b/src/main/resources/tessdata/configs/get.images @@ -0,0 +1 @@ +tessedit_write_images T diff --git a/src/main/resources/tessdata/configs/hocr b/src/main/resources/tessdata/configs/hocr new file mode 100644 index 0000000..5ab372e --- /dev/null +++ b/src/main/resources/tessdata/configs/hocr @@ -0,0 +1,2 @@ +tessedit_create_hocr 1 +hocr_font_info 0 diff --git a/src/main/resources/tessdata/configs/inter b/src/main/resources/tessdata/configs/inter new file mode 100644 index 0000000..252f1a1 --- /dev/null +++ b/src/main/resources/tessdata/configs/inter @@ -0,0 +1,2 @@ +interactive_display_mode T +tessedit_display_outwords T diff --git a/src/main/resources/tessdata/configs/kannada b/src/main/resources/tessdata/configs/kannada new file mode 100644 index 0000000..c6ac105 --- /dev/null +++ b/src/main/resources/tessdata/configs/kannada @@ -0,0 +1,4 @@ +textord_skewsmooth_offset 8 +textord_skewsmooth_offset2 8 +textord_merge_desc 0.5 +textord_no_rejects 1 diff --git a/src/main/resources/tessdata/configs/linebox b/src/main/resources/tessdata/configs/linebox new file mode 100644 index 0000000..bd9c114 --- /dev/null +++ b/src/main/resources/tessdata/configs/linebox @@ -0,0 +1,2 @@ +tessedit_resegment_from_line_boxes 1 +tessedit_make_boxes_from_boxes 1 diff --git a/src/main/resources/tessdata/configs/logfile b/src/main/resources/tessdata/configs/logfile new file mode 100644 index 0000000..a160f9b --- /dev/null +++ b/src/main/resources/tessdata/configs/logfile @@ -0,0 +1 @@ +debug_file tesseract.log diff --git a/src/main/resources/tessdata/configs/lstm.train b/src/main/resources/tessdata/configs/lstm.train new file mode 100644 index 0000000..5ff3772 --- /dev/null +++ b/src/main/resources/tessdata/configs/lstm.train @@ -0,0 +1,11 @@ +file_type .bl +textord_fast_pitch_test T +tessedit_zero_rejection T +tessedit_minimal_rejection F +tessedit_write_rep_codes F +edges_children_fix F +edges_childarea 0.65 +edges_boxarea 0.9 +tessedit_train_line_recognizer T +textord_no_rejects T +tessedit_init_config_only T diff --git a/src/main/resources/tessdata/configs/lstmbox b/src/main/resources/tessdata/configs/lstmbox new file mode 100644 index 0000000..a6f2ced --- /dev/null +++ b/src/main/resources/tessdata/configs/lstmbox @@ -0,0 +1 @@ +tessedit_create_lstmbox 1 diff --git a/src/main/resources/tessdata/configs/lstmdebug b/src/main/resources/tessdata/configs/lstmdebug new file mode 100644 index 0000000..3fa3dee --- /dev/null +++ b/src/main/resources/tessdata/configs/lstmdebug @@ -0,0 +1,4 @@ +stopper_debug_level 1 +classify_debug_level 1 +segsearch_debug_level 1 +language_model_debug_level 3 diff --git a/src/main/resources/tessdata/configs/makebox b/src/main/resources/tessdata/configs/makebox new file mode 100644 index 0000000..3d90ac2 --- /dev/null +++ b/src/main/resources/tessdata/configs/makebox @@ -0,0 +1 @@ +tessedit_create_boxfile 1 diff --git a/src/main/resources/tessdata/configs/page b/src/main/resources/tessdata/configs/page new file mode 100644 index 0000000..9928884 --- /dev/null +++ b/src/main/resources/tessdata/configs/page @@ -0,0 +1,3 @@ +tessedit_create_page_xml 1 +# page_xml_polygon 1 +# page_xml_level 0 diff --git a/src/main/resources/tessdata/configs/pdf b/src/main/resources/tessdata/configs/pdf new file mode 100644 index 0000000..59645d7 --- /dev/null +++ b/src/main/resources/tessdata/configs/pdf @@ -0,0 +1 @@ +tessedit_create_pdf 1 diff --git a/src/main/resources/tessdata/configs/quiet b/src/main/resources/tessdata/configs/quiet new file mode 100644 index 0000000..35b59a9 --- /dev/null +++ b/src/main/resources/tessdata/configs/quiet @@ -0,0 +1 @@ +debug_file /dev/null diff --git a/src/main/resources/tessdata/configs/rebox b/src/main/resources/tessdata/configs/rebox new file mode 100644 index 0000000..f8342b4 --- /dev/null +++ b/src/main/resources/tessdata/configs/rebox @@ -0,0 +1,2 @@ +tessedit_resegment_from_boxes 1 +tessedit_make_boxes_from_boxes 1 diff --git a/src/main/resources/tessdata/configs/strokewidth b/src/main/resources/tessdata/configs/strokewidth new file mode 100644 index 0000000..e95b592 --- /dev/null +++ b/src/main/resources/tessdata/configs/strokewidth @@ -0,0 +1,12 @@ +textord_show_blobs 0 +textord_debug_tabfind 3 +textord_tabfind_show_partitions 1 +textord_tabfind_show_initial_partitions 1 +textord_tabfind_show_columns 1 +textord_tabfind_show_blocks 1 +textord_tabfind_show_initialtabs 1 +textord_tabfind_show_finaltabs 1 +textord_tabfind_show_strokewidths 1 +textord_tabfind_show_vlines 0 +textord_tabfind_show_images 1 +tessedit_dump_pageseg_images 0 diff --git a/src/main/resources/tessdata/configs/tsv b/src/main/resources/tessdata/configs/tsv new file mode 100644 index 0000000..dc52478 --- /dev/null +++ b/src/main/resources/tessdata/configs/tsv @@ -0,0 +1 @@ +tessedit_create_tsv 1 diff --git a/src/main/resources/tessdata/configs/txt b/src/main/resources/tessdata/configs/txt new file mode 100644 index 0000000..a0cc952 --- /dev/null +++ b/src/main/resources/tessdata/configs/txt @@ -0,0 +1,3 @@ +# This config file should be used with other config files which create renderers. +# usage example: tesseract eurotext.tif eurotext txt hocr pdf +tessedit_create_txt 1 diff --git a/src/main/resources/tessdata/configs/unlv b/src/main/resources/tessdata/configs/unlv new file mode 100644 index 0000000..d2e22f5 --- /dev/null +++ b/src/main/resources/tessdata/configs/unlv @@ -0,0 +1,2 @@ +tessedit_write_unlv 1 +unlv_tilde_crunching T diff --git a/src/main/resources/tessdata/configs/wordstrbox b/src/main/resources/tessdata/configs/wordstrbox new file mode 100644 index 0000000..38cd41c --- /dev/null +++ b/src/main/resources/tessdata/configs/wordstrbox @@ -0,0 +1 @@ +tessedit_create_wordstrbox 1 -- cgit v1.2.3