AttributeError: 'Document' object has no attribute 'pageCount'

[memedacn]

解析pdf报错：
paddleocr --image_dir ./xxx.pdf --use_angle_cls true --use_gpu false --page_num 2
图片识别没问题：

[GreatV]

无法复现

paddleocr --image_dir ppstructure/docs/recovery/UnrealText.pdf --use_angle_cls true --use_gpu false --page_num 2
[2024/09/06 06:43:08] ppocr DEBUG: Namespace(help='==SUPPRESS==', use_gpu=False, use_xpu=False, use_npu=False, use_mlu=False, ir_optim=True, use_tensorrt=False, min_subgraph_size=15, precision='fp32', gpu_mem=500, gpu_id=0, image_dir='ppstructure/docs/recovery/UnrealText.pdf', page_num=2, det_algorithm='DB', det_model_dir='/home/greatx/.paddleocr/whl/det/ch/ch_PP-OCRv4_det_infer', det_limit_side_len=960, det_limit_type='max', det_box_type='quad', det_db_thresh=0.3, det_db_box_thresh=0.6, det_db_unclip_ratio=1.5, max_batch_size=10, use_dilation=False, det_db_score_mode='fast', det_east_score_thresh=0.8, det_east_cover_thresh=0.1, det_east_nms_thresh=0.2, det_sast_score_thresh=0.5, det_sast_nms_thresh=0.2, det_pse_thresh=0, det_pse_box_thresh=0.85, det_pse_min_area=16, det_pse_scale=1, scales=[8, 16, 32], alpha=1.0, beta=1.0, fourier_degree=5, rec_algorithm='SVTR_LCNet', rec_model_dir='/home/greatx/.paddleocr/whl/rec/ch/ch_PP-OCRv4_rec_infer', rec_image_inverse=True, rec_image_shape='3, 48, 320', rec_batch_num=6, max_text_length=25, rec_char_dict_path='/home/greatx/miniconda3/envs/ppocr/lib/python3.10/site-packages/paddleocr/ppocr/utils/ppocr_keys_v1.txt', use_space_char=True, vis_font_path='./doc/fonts/simfang.ttf', drop_score=0.5, e2e_algorithm='PGNet', e2e_model_dir=None, e2e_limit_side_len=768, e2e_limit_type='max', e2e_pgnet_score_thresh=0.5, e2e_char_dict_path='./ppocr/utils/ic15_dict.txt', e2e_pgnet_valid_set='totaltext', e2e_pgnet_mode='fast', use_angle_cls=True, cls_model_dir='/home/greatx/.paddleocr/whl/cls/ch_ppocr_mobile_v2.0_cls_infer', cls_image_shape='3, 48, 192', label_list=['0', '180'], cls_batch_num=6, cls_thresh=0.9, enable_mkldnn=False, cpu_threads=10, use_pdserving=False, warmup=False, sr_model_dir=None, sr_image_shape='3, 32, 128', sr_batch_num=1, draw_img_save_dir='./inference_results', save_crop_res=False, crop_res_save_dir='./output', use_mp=False, total_process_num=1, process_id=0, benchmark=False, save_log_path='./log_output/', show_log=True, use_onnx=False, return_word_box=False, output='./output', table_max_len=488, table_algorithm='TableAttn', table_model_dir=None, merge_no_span_structure=True, table_char_dict_path=None, layout_model_dir=None, layout_dict_path=None, layout_score_threshold=0.5, layout_nms_threshold=0.5, kie_algorithm='LayoutXLM', ser_model_dir=None, re_model_dir=None, use_visual_backbone=True, ser_dict_path='../train_data/XFUND/class_list_xfun.txt', ocr_order_method=None, mode='structure', image_orientation=False, layout=True, table=True, ocr=True, recovery=False, use_pdf2docx_api=False, invert=False, binarize=False, alphacolor=(255, 255, 255), lang='ch', det=True, rec=True, type='ocr', savefile=False, ocr_version='PP-OCRv4', structure_version='PP-StructureV2')
[2024/09/06 06:43:08] ppocr INFO: **********ppstructure/docs/recovery/UnrealText.pdf**********
[2024/09/06 06:43:08] ppocr DEBUG: dt_boxes num : 95, elapsed : 0.22710323333740234
[2024/09/06 06:43:08] ppocr DEBUG: cls num  : 95, elapsed : 0.06668782234191895
[2024/09/06 06:43:13] ppocr DEBUG: rec_res num  : 95, elapsed : 4.831298828125
[2024/09/06 06:43:13] ppocr DEBUG: dt_boxes num : 106, elapsed : 0.12230110168457031
[2024/09/06 06:43:14] ppocr DEBUG: cls num  : 106, elapsed : 0.06757664680480957
[2024/09/06 06:43:18] ppocr DEBUG: rec_res num  : 106, elapsed : 4.731462001800537
[2024/09/06 06:43:18] ppocr INFO: [[[125.0, 215.0], [1066.0, 215.0], [1066.0, 238.0], [125.0, 238.0]], ('UnrealText:SynthesizingRealisticSceneTextImagesfromtheUnrealWorld', 0.9962700605392456)]
[2024/09/06 06:43:18] ppocr INFO: [[[311.0, 287.0], [481.0, 291.0], [480.0, 319.0], [311.0, 315.0]], ('Shangbang Long', 0.9992293119430542)]
[2024/09/06 06:43:18] ppocr INFO: [[[715.0, 289.0], [817.0, 289.0], [817.0, 317.0], [715.0, 317.0]], ('Cong Yao', 0.9942947030067444)]
[2024/09/06 06:43:18] ppocr INFO: [[[261.0, 317.0], [531.0, 317.0], [531.0, 345.0], [261.0, 345.0]], ('Carnegie Mellon University', 0.9893227815628052)]
[2024/09/06 06:43:18] ppocr INFO: [[[600.0, 317.0], [926.0, 317.0], [926.0, 345.0], [600.0, 345.0]], ('Megvi (Face++) Technology Inc.', 0.9947630167007446)]
[2024/09/06 06:43:18] ppocr INFO: [[[294.0, 350.0], [501.0, 350.0], [501.0, 371.0], [294.0, 371.0]], ('[email protected]', 0.9976128935813904)]
[2024/09/06 06:43:18] ppocr INFO: [[[650.0, 350.0], [880.0, 348.0], [880.0, 371.0], [650.0, 373.0]], ('[email protected]', 0.9880282282829285)]
[2024/09/06 06:43:18] ppocr INFO: [[[38.0, 414.0], [63.0, 414.0], [63.0, 436.0], [38.0, 436.0]], ('O', 0.5061128735542297)]
[2024/09/06 06:43:18] ppocr INFO: [[[35.0, 429.0], [67.0, 429.0], [67.0, 457.0], [35.0, 457.0]], ('2', 0.9986442923545837)]
[2024/09/06 06:43:18] ppocr INFO: [[[34.0, 473.0], [64.0, 468.0], [67.0, 490.0], [37.0, 494.0]], ('2', 0.9995098114013672)]
[2024/09/06 06:43:18] ppocr INFO: [[[236.0, 499.0], [270.0, 504.0], [267.0, 524.0], [234.0, 519.0]], ('fon', 0.8305749893188477)]
[2024/09/06 06:43:18] ppocr INFO: [[[37.0, 541.0], [68.0, 541.0], [68.0, 571.0], [37.0, 571.0]], ('A', 0.9159913063049316)]
[2024/09/06 06:43:18] ppocr INFO: [[[826.0, 532.0], [899.0, 528.0], [900.0, 552.0], [828.0, 556.0]], ('ali erat', 0.8404781818389893)]
[2024/09/06 06:43:18] ppocr INFO: [[[141.0, 551.0], [171.0, 551.0], [171.0, 569.0], [141.0, 569.0]], ('PSz', 0.7466688752174377)]
[2024/09/06 06:43:18] ppocr INFO: [[[828.0, 556.0], [901.0, 556.0], [901.0, 587.0], [828.0, 587.0]], ('re der', 0.9081944823265076)]
[2024/09/06 06:43:18] ppocr INFO: [[[933.0, 549.0], [958.0, 549.0], [958.0, 563.0], [933.0, 563.0]], ('$BP', 0.9890866279602051)]
[2024/09/06 06:43:18] ppocr INFO: [[[36.0, 576.0], [70.0, 582.0], [65.0, 608.0], [32.0, 603.0]], ('8', 0.9981963038444519)]
[2024/09/06 06:43:18] ppocr INFO: [[[838.0, 601.0], [914.0, 597.0], [915.0, 615.0], [839.0, 619.0]], ('piorrru', 0.5357216596603394)]
[2024/09/06 06:43:18] ppocr INFO: [[[191.0, 648.0], [328.0, 648.0], [328.0, 670.0], [191.0, 670.0]], ('Lighting & Shadows', 0.9710700511932373)]
[2024/09/06 06:43:18] ppocr INFO: [[[526.0, 648.0], [664.0, 648.0], [664.0, 670.0], [526.0, 670.0]], ('Suitable Text Region', 0.9951587915420532)]
[2024/09/06 06:43:18] ppocr INFO: [[[895.0, 650.0], [961.0, 650.0], [961.0, 668.0], [895.0, 668.0]], ('Occlusion', 0.9946575164794922)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 670.0], [1094.0, 673.0], [1094.0, 701.0], [95.0, 698.0]], ('Figure 1: Demonstration of the proposed UnrealText synthesis engine, which achieves photo-realistic lighting conditions,', 0.9844685792922974)]
[2024/09/06 06:43:18] ppocr INFO: [[[45.0, 723.0], [67.0, 723.0], [67.0, 738.0], [45.0, 738.0]], ('S', 0.9632604718208313)]
[2024/09/06 06:43:18] ppocr INFO: [[[43.0, 734.0], [68.0, 734.0], [68.0, 756.0], [43.0, 756.0]], ('C', 0.610887885093689)]
[2024/09/06 06:43:18] ppocr INFO: [[[291.0, 739.0], [384.0, 739.0], [384.0, 762.0], [291.0, 762.0]], ('Abstract', 0.9991759061813354)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 741.0], [1091.0, 741.0], [1091.0, 762.0], [617.0, 762.0]], ('els are data-thirsty,and it is expensive and sometimes dif-', 0.9581263661384583)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 761.0], [1093.0, 762.0], [1093.0, 790.0], [614.0, 789.0]], ('ficult, if not impossible, to collect enough data. More-', 0.9830140471458435)]
[2024/09/06 06:43:18] ppocr INFO: [[[121.0, 785.0], [574.0, 787.0], [574.0, 810.0], [121.0, 808.0]], ('Syntheticdatahasbeenacriticaltoolfortrainingscene', 0.9984138607978821)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 789.0], [1093.0, 789.0], [1093.0, 812.0], [615.0, 812.0]], ('over, the various applications, from traffic sign reading in', 0.9964367747306824)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 812.0], [574.0, 812.0], [574.0, 833.0], [98.0, 833.0]], ('textdetectionandrecognitionmodels.Ontheonehand,', 0.992158830165863)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 812.0], [1093.0, 812.0], [1093.0, 835.0], [617.0, 835.0]], ('autonomous vehicles to instant translation, require a large', 0.9815146327018738)]
[2024/09/06 06:43:18] ppocr INFO: [[[37.0, 827.0], [65.0, 827.0], [65.0, 848.0], [37.0, 848.0]], ('8', 0.9992054104804993)]
[2024/09/06 06:43:18] ppocr INFO: [[[96.0, 833.0], [575.0, 833.0], [575.0, 861.0], [96.0, 861.0]], ('synthetic word images have proven tobe a successful sub-', 0.9793184995651245)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 833.0], [1093.0, 833.0], [1093.0, 861.0], [614.0, 861.0]], ('amount of data specifically for each domain, further es-', 0.9856284260749817)]
[2024/09/06 06:43:18] ppocr INFO: [[[38.0, 866.0], [63.0, 866.0], [63.0, 886.0], [38.0, 886.0]], ('6', 0.9971768856048584)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 860.0], [577.0, 860.0], [577.0, 883.0], [98.0, 883.0]], ('stituteforrealimagesintrainingscenetextrecognizers.On', 0.9984899759292603)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 858.0], [1089.0, 858.0], [1089.0, 881.0], [614.0, 881.0]], ('calating this issue. Therefore, synthetic data and synthe-', 0.9836729764938354)]
[2024/09/06 06:43:18] ppocr INFO: [[[97.0, 881.0], [575.0, 883.0], [575.0, 906.0], [96.0, 904.0]], ('theotherhand,however,scenetextdetectors stillheavily', 0.982474684715271)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 884.0], [1091.0, 884.0], [1091.0, 908.0], [615.0, 908.0]], ('sis algorithms are important for scene text tasks. Further-', 0.9945923686027527)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 908.0], [577.0, 908.0], [577.0, 931.0], [100.0, 931.0]], ('relyonalargeamountofmanuallyannotatedreal-world', 0.9988702535629272)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 908.0], [1093.0, 908.0], [1093.0, 931.0], [615.0, 931.0]], ('more, synthetic data can provide detailed annotations, such', 0.9902983903884888)]
[2024/09/06 06:43:18] ppocr INFO: [[[35.0, 939.0], [67.0, 939.0], [67.0, 959.0], [35.0, 959.0]], ('3', 0.9980735778808594)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 932.0], [575.0, 932.0], [575.0, 955.0], [100.0, 955.0]], ('images,which are expensive.In this paper,we introduce', 0.9575583338737488)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 932.0], [1094.0, 932.0], [1094.0, 955.0], [615.0, 955.0]], ('as character-level or even pixel-level ground truths that are', 0.9933388233184814)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 955.0], [574.0, 955.0], [574.0, 978.0], [100.0, 978.0]], ('UnrealText,anefficientimagesynthesis methodthatren-', 0.987811803817749)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 955.0], [911.0, 955.0], [911.0, 978.0], [615.0, 978.0]], ('rare for real images due to high cost.', 0.9884402751922607)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 980.0], [574.0, 980.0], [574.0, 1003.0], [98.0, 1003.0]], ('ders realisticimagesvia a 3Dgraphics engine.3Dsyn-', 0.9644061923027039)]
[2024/09/06 06:43:18] ppocr INFO: [[[639.0, 985.0], [1089.0, 985.0], [1089.0, 1008.0], [639.0, 1008.0]], ('Currently, there exist several synthesis algorithms [46,', 0.9953686594963074)]
[2024/09/06 06:43:18] ppocr INFO: [[[37.0, 1000.0], [67.0, 1000.0], [67.0, 1018.0], [37.0, 1018.0]], ('2', 0.999278724193573)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 1000.0], [577.0, 1002.0], [577.0, 1030.0], [95.0, 1028.0]], ('thetic engine provides realistic appearance by rendering', 0.9842777252197266)]
[2024/09/06 06:43:18] ppocr INFO: [[[620.0, 1010.0], [1093.0, 1010.0], [1093.0, 1033.0], [620.0, 1033.0]], ('10,6, 50] that have proven beneficial. Especially, in scene', 0.9881604909896851)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 1028.0], [572.0, 1028.0], [572.0, 1051.0], [100.0, 1051.0]], ('scene andtext as awhole,and allowsforbettertextre-', 0.9560279846191406)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1031.0], [1094.0, 1030.0], [1094.0, 1058.0], [614.0, 1059.0]], ('text recognition, training on synthetic data [10, 6] alone', 0.9911512732505798)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1051.0], [574.0, 1051.0], [574.0, 1074.0], [98.0, 1074.0]], ('gionproposalswith accesstoprecisesceneinformation,', 0.9935410022735596)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1058.0], [1091.0, 1058.0], [1091.0, 1081.0], [615.0, 1081.0]], ('has become a widely accepted standard practice. Some re-', 0.9878241419792175)]
[2024/09/06 06:43:18] ppocr INFO: [[[93.0, 1073.0], [575.0, 1071.0], [575.0, 1099.0], [93.0, 1101.0]], ('e.g.normal and even object meshes.The comprehensive', 0.9944592714309692)]
[2024/09/06 06:43:18] ppocr INFO: [[[45.0, 1089.0], [65.0, 1089.0], [65.0, 1119.0], [45.0, 1119.0]], ('ar', 0.9955083131790161)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1081.0], [1091.0, 1081.0], [1091.0, 1104.0], [615.0, 1104.0]], ('searchers that attempt training on both synthetic and real', 0.9810498952865601)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1099.0], [574.0, 1099.0], [574.0, 1122.0], [98.0, 1122.0]], ('experimentsverifyits effectiveness onbothscenetext de-', 0.9801911115646362)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1104.0], [1091.0, 1104.0], [1091.0, 1127.0], [615.0, 1127.0]], ('data only report marginal improvements [15, 20] on most', 0.9832894802093506)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1119.0], [577.0, 1119.0], [577.0, 1147.0], [98.0, 1147.0]], ('tection and recognition.We also generate a multilingual', 0.9661106467247009)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 1130.0], [1091.0, 1130.0], [1091.0, 1153.0], [617.0, 1153.0]], ('datasets. Mixing synthetic and real data is only improving', 0.984853982925415)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 1147.0], [574.0, 1147.0], [574.0, 1170.0], [100.0, 1170.0]], ('versionforfutureresearchintomultilingualscenetextde-', 0.9976158738136292)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1153.0], [1093.0, 1153.0], [1093.0, 1176.0], [615.0, 1176.0]], ('performance on a few difficult cases that are not yet well', 0.984887421131134)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1170.0], [575.0, 1170.0], [575.0, 1193.0], [98.0, 1193.0]], ('tection andrecognition.Additionally,were-annotatescene', 0.9803580045700073)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1178.0], [1091.0, 1178.0], [1091.0, 1201.0], [615.0, 1201.0]], ('covered by existing synthetic datasets, such as seriously', 0.9890692234039307)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1195.0], [574.0, 1195.0], [574.0, 1218.0], [98.0, 1218.0]], ('textrecognitiondatasetsin a case-sensitivewayandin-', 0.9819285273551941)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1200.0], [1094.0, 1200.0], [1094.0, 1228.0], [614.0, 1228.0]], ('blurred or curved text. This is reasonable, since cropped', 0.9922405481338501)]
[2024/09/06 06:43:18] ppocr INFO: [[[97.0, 1214.0], [575.0, 1216.0], [575.0, 1244.0], [96.0, 1242.0]], ('clude punctuation marks for more comprehensive evalua-', 0.9768950939178467)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1226.0], [1093.0, 1226.0], [1093.0, 1249.0], [615.0, 1249.0]], ('text images have much simpler background, and synthetic', 0.9979594349861145)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1242.0], [575.0, 1242.0], [575.0, 1266.0], [98.0, 1266.0]], ('tions.The codeand thegenerated datasets arereleased at:', 0.9596098065376282)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1267.0], [399.0, 1267.0], [399.0, 1290.0], [98.0, 1290.0]], ('https://jyouhou.github.io/UnrealText/', 0.9850106239318848)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1274.0], [1094.0, 1274.0], [1094.0, 1297.0], [615.0, 1297.0]], ('sity of backgrounds, fonts, and lighting conditions, as well', 0.9813407063484192)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1297.0], [948.0, 1297.0], [948.0, 1320.0], [615.0, 1320.0]], ('as thousands of times more data samples.', 0.9794551730155945)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1312.0], [256.0, 1312.0], [256.0, 1335.0], [98.0, 1335.0]], ('1. Introduction', 0.9978073835372925)]
[2024/09/06 06:43:18] ppocr INFO: [[[635.0, 1323.0], [1094.0, 1325.0], [1094.0, 1353.0], [635.0, 1351.0]], (' On the contrary, however, scene text detection is still', 0.970913827419281)]
[2024/09/06 06:43:18] ppocr INFO: [[[121.0, 1351.0], [572.0, 1351.0], [572.0, 1374.0], [121.0, 1374.0]], ('With the resurgence of neural networks, the past few', 0.9657694101333618)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1351.0], [1091.0, 1351.0], [1091.0, 1374.0], [615.0, 1374.0]], ('heavily dependent on real-world data. Synthetic data [6, 50]', 0.9801923036575317)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1376.0], [577.0, 1376.0], [577.0, 1399.0], [98.0, 1399.0]], ('years have witnessed significant progress in the field of', 0.9765846729278564)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1376.0], [1089.0, 1376.0], [1089.0, 1399.0], [614.0, 1399.0]], ('plays a less significant role, and only brings marginal im-', 0.9755415320396423)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 1399.0], [574.0, 1399.0], [574.0, 1422.0], [100.0, 1422.0]], ('scene text detection and recognition. However, these mod-', 0.9830067753791809)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 1398.0], [1094.0, 1396.0], [1094.0, 1424.0], [612.0, 1426.0]], ('provements. Existing synthesizers for scene text detec-', 0.9788098335266113)]
[2024/09/06 06:43:18] ppocr INFO: [[[589.0, 1468.0], [602.0, 1468.0], [602.0, 1483.0], [589.0, 1483.0]], ('1', 0.9976250529289246)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 148.0], [572.0, 148.0], [572.0, 172.0], [98.0, 172.0]], ('tion follow the same paradigm. First, they analyze back-', 0.987568736076355)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 148.0], [1091.0, 148.0], [1091.0, 172.0], [615.0, 172.0]], ('to carry out comprehensive evaluations, and tend to over-', 0.9843573570251465)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 170.0], [577.0, 168.0], [577.0, 196.0], [95.0, 198.0]], ('ground images, e.g. by performing semantic segmentation', 0.9905786514282227)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 170.0], [1093.0, 170.0], [1093.0, 198.0], [614.0, 198.0]], ('estimate the progress of scene text recognition algorithms.', 0.9822967648506165)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 191.0], [575.0, 195.0], [575.0, 223.0], [95.0, 219.0]], ('and depth estimation using off-the-shelf models. Then, po-', 0.970655620098114)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 196.0], [1091.0, 196.0], [1091.0, 219.0], [617.0, 219.0]], ('To address this issue, we re-annotate these datasets to in-', 0.9929389357566833)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 218.0], [574.0, 218.0], [574.0, 241.0], [98.0, 241.0]], ('tential locations for text embedding are extracted from the', 0.9900808930397034)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 218.0], [1091.0, 218.0], [1091.0, 241.0], [615.0, 241.0]], ('clude both upper-case and lower-case characters,digits,', 0.9842342734336853)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 244.0], [575.0, 244.0], [575.0, 267.0], [98.0, 267.0]], ('segmented regions. Finally, text images (foregrounds) are', 0.9911876916885376)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 241.0], [1094.0, 243.0], [1094.0, 271.0], [612.0, 269.0]], ('punctuation marks, and spaces if there are any. We urge', 0.9767991304397583)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 267.0], [574.0, 267.0], [574.0, 290.0], [98.0, 290.0]], ('blended into the background images, with perceptive trans-', 0.9933977723121643)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 267.0], [1093.0, 267.0], [1093.0, 290.0], [615.0, 290.0]], ('researchers to use the new annotations and evaluate in such', 0.9890763163566589)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 290.0], [575.0, 290.0], [575.0, 314.0], [98.0, 314.0]], ('formation inferred from estimated depth. However, the', 0.9801830053329468)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 290.0], [1091.0, 290.0], [1091.0, 314.0], [615.0, 314.0]], ('a full-symbol mode for better understanding of the advan-', 0.985104501247406)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 315.0], [575.0, 315.0], [575.0, 338.0], [98.0, 338.0]], ('analysis of background images with off-the-shelf models', 0.9955949783325195)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 315.0], [998.0, 315.0], [998.0, 338.0], [614.0, 338.0]], ('tages and disadvantages of different algorithms.', 0.9987747073173523)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 335.0], [577.0, 337.0], [577.0, 365.0], [95.0, 363.0]], ('may be rough and imprecise. The errors further propagate', 0.9909663796424866)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 363.0], [575.0, 363.0], [575.0, 386.0], [98.0, 386.0]], ('to text proposal modules and result in text being embedded', 0.9972422122955322)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 365.0], [785.0, 365.0], [785.0, 388.0], [617.0, 388.0]], ('2.Related Work', 0.9692772030830383)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 383.0], [577.0, 384.0], [577.0, 413.0], [95.0, 411.0]], (' onto unsuitable locations. Moreover, the text embedding', 0.9776049256324768)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 401.0], [820.0, 404.0], [820.0, 432.0], [613.0, 429.0]], ('2.1. Synthetic Images', 0.9938879609107971)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 411.0], [575.0, 411.0], [575.0, 434.0], [98.0, 434.0]], ('process is ignorant of the overall image conditions such as', 0.9934335350990295)]
[2024/09/06 06:43:18] ppocr INFO: [[[97.0, 432.0], [575.0, 434.0], [575.0, 457.0], [96.0, 455.0]], ('illumination and occlusions of the scene.These two factors', 0.9771759510040283)]
[2024/09/06 06:43:18] ppocr INFO: [[[639.0, 441.0], [1091.0, 442.0], [1091.0, 470.0], [639.0, 469.0]], ('The synthesis of photo-realistic datasets has been a pop-', 0.9803728461265564)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 454.0], [577.0, 457.0], [577.0, 485.0], [95.0, 482.0]], ('make text instances outstanding from backgrounds, leading', 0.9974026679992676)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 469.0], [1089.0, 469.0], [1089.0, 492.0], [615.0, 492.0]], ('ular topic, since they provide detailed ground-truth annota-', 0.9846197366714478)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 483.0], [446.0, 483.0], [446.0, 507.0], [98.0, 507.0]], ('to a gap between synthetic and real images.', 0.9939976334571838)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 493.0], [1091.0, 493.0], [1091.0, 516.0], [615.0, 516.0]], ('tions at multiple granularity, and cost less than manual an-', 0.9868430495262146)]
[2024/09/06 06:43:18] ppocr INFO: [[[121.0, 510.0], [574.0, 510.0], [574.0, 533.0], [121.0, 533.0]], ('In this paper, we propose a synthetic engine that syn-', 0.9927743077278137)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 518.0], [1094.0, 518.0], [1094.0, 540.0], [615.0, 540.0]], ('notations. In scene text detection and recognition, the use of', 0.99276202917099)]
[2024/09/06 06:43:18] ppocr INFO: [[[93.0, 531.0], [530.0, 530.0], [531.0, 558.0], [93.0, 559.0]], ('thesizes scene text images from 3D virtual world.', 0.9793897271156311)]
[2024/09/06 06:43:18] ppocr INFO: [[[521.0, 536.0], [574.0, 536.0], [574.0, 554.0], [521.0, 554.0]], ('The', 0.9997890591621399)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 540.0], [1093.0, 540.0], [1093.0, 563.0], [615.0, 563.0]], ('synthetic datasets has become a standard practice.For scene', 0.9867520928382874)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 556.0], [579.0, 554.0], [579.0, 582.0], [95.0, 584.0]], ('proposed engine is based on the famous Unreal Engine 4', 0.994499683380127)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 564.0], [1091.0, 564.0], [1091.0, 587.0], [615.0, 587.0]], ('text recognition, where images contain only one word, syn-', 0.9852733612060547)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 582.0], [574.0, 582.0], [574.0, 606.0], [98.0, 606.0]], ('(UE4), and is therefore named as UnrealText. Specifically,', 0.9750241041183472)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 587.0], [1091.0, 587.0], [1091.0, 610.0], [615.0, 610.0]], ('thetic images are rendered through several steps [46, 10],', 0.9939011931419373)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 604.0], [577.0, 602.0], [577.0, 630.0], [95.0, 632.0]], ('text instances are regarded as planar polygon meshes with', 0.9829094409942627)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 612.0], [1091.0, 612.0], [1091.0, 635.0], [615.0, 635.0]], ('including font rendering, coloring, homography transfor-', 0.9919406175613403)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 630.0], [575.0, 630.0], [575.0, 653.0], [98.0, 653.0]], ('text foregrounds loaded as texture. These meshes are placed', 0.9962610602378845)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 634.0], [1094.0, 634.0], [1094.0, 662.0], [614.0, 662.0]], ('mation, and background blending. Later, GANs [5] are', 0.9922314882278442)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 653.0], [575.0, 653.0], [575.0, 676.0], [98.0, 676.0]], ('in suitable positions in 3D world, and rendered together', 0.9922076463699341)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 660.0], [1093.0, 660.0], [1093.0, 683.0], [617.0, 683.0]], ('incorporated to maintain style consistency for implanted', 0.9874199032783508)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 678.0], [309.0, 678.0], [309.0, 700.0], [98.0, 700.0]], ('with the scene as a whole.', 0.9800056219100952)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 680.0], [1093.0, 683.0], [1092.0, 710.0], [614.0, 706.0]], ('text [51], but it is only for single-word images. As a re-', 0.9739071130752563)]
[2024/09/06 06:43:18] ppocr INFO: [[[123.0, 705.0], [572.0, 705.0], [572.0, 728.0], [123.0, 728.0]], ('As shown in Fig. 1, the proposed synthesis engine, by', 0.9910547733306885)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 708.0], [1093.0, 708.0], [1093.0, 731.0], [615.0, 731.0]], ('sult of these progresses, synthetic data alone are enough to', 0.994909942150116)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 729.0], [574.0, 729.0], [574.0, 752.0], [98.0, 752.0]], ('its very nature, enjoys the following advantages over pre-', 0.9947152137756348)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 733.0], [880.0, 733.0], [880.0, 756.0], [615.0, 756.0]], ('train state-of-the-art recognizers.', 0.9836069345474243)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 752.0], [575.0, 752.0], [575.0, 776.0], [98.0, 776.0]], ('vious methods: (1) Text and scenes are rendered together,', 0.9879263043403625)]
[2024/09/06 06:43:18] ppocr INFO: [[[639.0, 754.0], [1093.0, 757.0], [1093.0, 781.0], [639.0, 777.0]], ('To train scene text detectors, SynthText [6] proposes to', 0.9919214248657227)]
[2024/09/06 06:43:18] ppocr INFO: [[[96.0, 776.0], [575.0, 776.0], [575.0, 804.0], [96.0, 804.0]], ('achieving realistic visual effects, e.g. illumination, occlu-', 0.9955040812492371)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 782.0], [1093.0, 782.0], [1093.0, 805.0], [615.0, 805.0]], ('generate synthetic data by printing text on background im-', 0.9909249544143677)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 800.0], [575.0, 800.0], [575.0, 823.0], [98.0, 823.0]], ('sion, and perspective transformation. (2) The method has', 0.9886850714683533)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 804.0], [1094.0, 800.0], [1094.0, 828.0], [612.0, 832.0]], (' ages. It first analyzes images with off-the-shelf models, and', 0.9571012854576111)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 825.0], [575.0, 825.0], [575.0, 848.0], [98.0, 848.0]], ('access to precise scene information, e.g. normal, depth, and', 0.9852426052093506)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 828.0], [1089.0, 828.0], [1089.0, 851.0], [617.0, 851.0]], ('search suitable text regions on semantically consistent re-', 0.981018602848053)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 848.0], [574.0, 848.0], [574.0, 871.0], [98.0, 871.0]], ('object meshes, and therefore can generate better text region', 0.976837694644928)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 853.0], [1093.0, 853.0], [1093.0, 876.0], [617.0, 876.0]], ('gions. Text are implanted with perspective transformation', 0.993736743927002)]
[2024/09/06 06:43:18] ppocr INFO: [[[93.0, 871.0], [557.0, 870.0], [557.0, 898.0], [93.0, 899.0]], (' proposals. These aspects are crucial in training detectors.', 0.9905708432197571)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 873.0], [1093.0, 875.0], [1093.0, 903.0], [612.0, 901.0]], ('based on estimated depth. To maintain semantic coherency,', 0.9921225309371948)]
[2024/09/06 06:43:18] ppocr INFO: [[[118.0, 896.0], [577.0, 898.0], [577.0, 926.0], [118.0, 924.0]], ('To further exploit the potential of UnrealText, we design', 0.9809216260910034)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 899.0], [1091.0, 899.0], [1091.0, 922.0], [615.0, 922.0]], ('VISD [50] proposes to use semantic segmentation to filter', 0.9682649374008179)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 922.0], [575.0, 922.0], [575.0, 945.0], [98.0, 945.0]], ('three key components: (1) A view finding algorithm that', 0.9692460894584656)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 926.0], [1093.0, 926.0], [1093.0, 949.0], [615.0, 949.0]], ('out unreasonable surfaces such as human faces. They also', 0.9837173819541931)]
[2024/09/06 06:43:18] ppocr INFO: [[[100.0, 949.0], [575.0, 949.0], [575.0, 972.0], [100.0, 972.0]], ('explores the virtual scenes and generates camera viewpoints', 0.9931870102882385)]
[2024/09/06 06:43:18] ppocr INFO: [[[617.0, 950.0], [1093.0, 950.0], [1093.0, 972.0], [617.0, 972.0]], ('adopt an adaptive coloring scheme to fit the text into the', 0.974525511264801)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 972.0], [572.0, 972.0], [572.0, 995.0], [98.0, 995.0]], ('to obtain more diverse and natural backgrounds. (2) An en-', 0.992268443107605)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 972.0], [1091.0, 972.0], [1091.0, 995.0], [615.0, 995.0]], ('artistic style of backgrounds. However, without consider-', 0.9939703941345215)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 997.0], [575.0, 997.0], [575.0, 1020.0], [98.0, 1020.0]], ('vironment randomization module that changes the lighting', 0.9969351887702942)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 997.0], [1093.0, 997.0], [1093.0, 1020.0], [615.0, 1020.0]], ('ing the scene as a whole, these methods fail to render text', 0.9814044237136841)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1020.0], [542.0, 1020.0], [542.0, 1043.0], [98.0, 1043.0]], ('conditions regularly, to simulate real-world variations.', 0.995219349861145)]
[2024/09/06 06:43:18] ppocr INFO: [[[534.0, 1021.0], [574.0, 1021.0], [574.0, 1041.0], [534.0, 1041.0]], ('.(3)', 0.8634718656539917)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1020.0], [1093.0, 1020.0], [1093.0, 1043.0], [615.0, 1043.0]], ('instances in a photo-realistic way, and text instances are too', 0.9914546608924866)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1044.0], [575.0, 1044.0], [575.0, 1068.0], [98.0, 1068.0]], ('A mesh-based text region generation method that finds suit-', 0.9925909042358398)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 1040.0], [1093.0, 1040.0], [1093.0, 1068.0], [612.0, 1068.0]], (' outstanding from backgrounds. So far, the training of de-', 0.9921371340751648)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 1064.0], [494.0, 1066.0], [494.0, 1094.0], [95.0, 1092.0]], (' able positions for text by probing the 3D meshes.', 0.9894159436225891)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 1064.0], [945.0, 1066.0], [945.0, 1094.0], [612.0, 1092.0]], ('tectors still relies heavily on real images.', 0.9761550426483154)]
[2024/09/06 06:43:18] ppocr INFO: [[[118.0, 1092.0], [575.0, 1091.0], [575.0, 1119.0], [118.0, 1120.0]], ('The contributions of this paper are summarized as fol-', 0.9971570372581482)]
[2024/09/06 06:43:18] ppocr INFO: [[[640.0, 1094.0], [1093.0, 1094.0], [1093.0, 1117.0], [640.0, 1117.0]], ('Although GANs and other learning-based methods have', 0.996195912361145)]
[2024/09/06 06:43:18] ppocr INFO: [[[96.0, 1117.0], [575.0, 1117.0], [575.0, 1145.0], [96.0, 1145.0]], ('lows: (1) We propose a brand-new scene text image syn-', 0.9715551733970642)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1115.0], [1091.0, 1115.0], [1091.0, 1138.0], [615.0, 1138.0]], ('also shown great potential in generating realistic im-', 0.9574754238128662)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1143.0], [575.0, 1143.0], [575.0, 1167.0], [98.0, 1167.0]], ('thesis engine that renders images from 3D world, which is', 0.9838658571243286)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1138.0], [1094.0, 1138.0], [1094.0, 1167.0], [614.0, 1167.0]], ('ages [48, 17, 12], the generation of scene text images still', 0.9831877946853638)]
[2024/09/06 06:43:18] ppocr INFO: [[[96.0, 1165.0], [577.0, 1165.0], [577.0, 1193.0], [96.0, 1193.0]], ('entirely different from previous approaches that embed text', 0.9871304631233215)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1165.0], [1091.0, 1165.0], [1091.0, 1188.0], [615.0, 1188.0]], ('require a large amount of manually labeled data [51]. Fur-', 0.9811139106750488)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1190.0], [572.0, 1190.0], [572.0, 1213.0], [98.0, 1213.0]], ('on 2D background images, termed as UnrealText. The pro-', 0.991174042224884)]
[2024/09/06 06:43:18] ppocr INFO: [[[612.0, 1185.0], [1094.0, 1186.0], [1094.0, 1214.0], [612.0, 1213.0]], ('thermore, such data are sometimes not easy to collect, es-', 0.9920262694358826)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 1213.0], [577.0, 1211.0], [577.0, 1239.0], [95.0, 1241.0]], (' posed engine achieves realistic rendering effects and high', 0.9874371290206909)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1213.0], [1018.0, 1214.0], [1018.0, 1238.0], [614.0, 1236.0]], ('pecially for cases such as low resource languages.', 0.9834662079811096)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1238.0], [570.0, 1238.0], [570.0, 1261.0], [98.0, 1261.0]], ('scalability. (2) With the proposed techniques, the synthe-', 0.9654174447059631)]
[2024/09/06 06:43:18] ppocr INFO: [[[640.0, 1239.0], [1089.0, 1239.0], [1089.0, 1262.0], [640.0, 1262.0]], ('More recently, synthesizing images with 3D graph-', 0.9684094786643982)]
[2024/09/06 06:43:18] ppocr INFO: [[[95.0, 1261.0], [574.0, 1259.0], [574.0, 1287.0], [95.0, 1289.0]], ('sis engine improves the performance of detectors and rec-', 0.9895493984222412)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1261.0], [1093.0, 1261.0], [1093.0, 1289.0], [614.0, 1289.0]], ('ics engine has become popular in several fields, in-', 0.9903081059455872)]
[2024/09/06 06:43:18] ppocr INFO: [[[96.0, 1284.0], [575.0, 1284.0], [575.0, 1312.0], [96.0, 1312.0]], ('ognizers significantly. (3) We also generate a large scale', 0.9729388952255249)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1285.0], [1091.0, 1285.0], [1091.0, 1308.0], [615.0, 1308.0]], ('cluding human pose estimation [43], scene understand-', 0.9837570786476135)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1310.0], [574.0, 1310.0], [574.0, 1333.0], [98.0, 1333.0]], ('multilingual scene text dataset that will aid further research.', 0.9774734973907471)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1310.0], [1089.0, 1310.0], [1089.0, 1333.0], [615.0, 1333.0]], ('ing/segmentation [28,24, 33, 35, 37], and object detec-', 0.9710452556610107)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1335.0], [575.0, 1335.0], [575.0, 1358.0], [98.0, 1358.0]], ('(4) Additionally, we notice that many of the popular scene', 0.9869992733001709)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1333.0], [1093.0, 1333.0], [1093.0, 1356.0], [615.0, 1356.0]], ('tion [29,42, 8]. However, these methods either consider', 0.9837771058082581)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1360.0], [574.0, 1360.0], [574.0, 1381.0], [98.0, 1381.0]], ('text recognition datasets are only annotated in an incom-', 0.9763432145118713)]
[2024/09/06 06:43:18] ppocr INFO: [[[614.0, 1356.0], [1093.0, 1356.0], [1093.0, 1384.0], [614.0, 1384.0]], ('simplistic cases, e.g. rendering 3D objects on top of static', 0.9775846004486084)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1383.0], [574.0, 1383.0], [574.0, 1406.0], [98.0, 1406.0]], ('plete way, providing only case-insensitive word annota-', 0.9897478818893433)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1383.0], [1091.0, 1383.0], [1091.0, 1406.0], [615.0, 1406.0]], ('background images [29, 43] and randomly arranging scenes', 0.9892158508300781)]
[2024/09/06 06:43:18] ppocr INFO: [[[98.0, 1406.0], [575.0, 1406.0], [575.0, 1429.0], [98.0, 1429.0]], ('tions. With such limited annotations, researchers are unable', 0.989226222038269)]
[2024/09/06 06:43:18] ppocr INFO: [[[615.0, 1406.0], [1089.0, 1406.0], [1089.0, 1429.0], [615.0, 1429.0]], ('filled with objects [28, 24, 35, 8], or passively use off-the-', 0.9829160571098328)]