在上一篇文章中,我们成功串联了文本检测、方向分类、文本识别模型,搭建了一个基础的OCR文本识别服务。然而,针对复杂的PDF文档,仅靠基础的OCR功能还远远不够。为了实现全面的PDF文档解析,尤其是对于表格和版面内容的识别,还需要串联更多的目标检测模型,如表格识别和版面分析模型。
本篇文章将基于前两篇的内容,深入讲解如何将文本、表格、版面三个模型进行串联部署。通过对这些模型的串联,不仅可以处理文本,还能解析复杂的表格和多种布局格式的文档,使OCR系统具有商用水平。
在实践中,串联的文档解析系统能够处理各种高难度场景,如发票、合同、金融报表等包含复杂结构的文档。不仅支持传统的文本识别,还能够识别数学公式、表格、图表等内容,并且可以将解析结果输出为JSON、Markdown、HTML等多种格式,适应不同应用场景。
通过这些模型的串联和部署,用户可以构建一个完整的OCR文档解析服务,并支持企业自定义的模型训练和部署,进一步拓展为云平台API服务。这意味着企业可以根据自身需求定制模型并将其发布为服务,供用户调用,提升整体业务效率和文档处理能力。
Windows 环境下安装模型运行环境以及发布OCR文本识别服务的具体步骤和说明,请参考之前的文章,链接如下:
| Windows下OCR模型运行环境安装 |
| Windows安装PaddleHub |
PaddleOCR通过PaddleHub Serving发布串联的文本、表格、版面识别模型,利用structure_system模块实现文档解析的整体流程。具体过程如下:
工作原理:
文本检测模型:识别图像中的文字块位置,输出对应的检测框。
方向分类模型:矫正文字检测框的方向,确保字符的正确方向排列。
文本识别模型:从矫正后的框中提取文字,并输出识别结果。
表格识别:对于表格内容,模型会分析表格的结构和单元格,提取出相应的文本和位置关系。
texttitlefigurefigure_captiontabletable_captionheaderfooterreferenceequation
模型对于每个word都会预测出一个标签,相同标签的word聚合成一个实体(如titles、text、header等),而实体的外接框为实体中所有word的外接框。
可视化结果如下(不同颜色表示不同实体,如红色表示标题,青绿表示表格,蓝色表示页眉,绿色表示脚注等):
deploy/hubserving/└─ocr_cls文本方向分类模块服务包└─ocr_det文本检测模块服务包└─ocr_rec文本识别模块服务包└─ocr_system文本检测+文本方向分类+文本识别串联服务包└─structure_layout版面分析服务包└─structure_table表格识别服务包└─structure_systemPP-Structure服务包└─kie_ser关键信息抽取-SER服务包└─kie_ser_re关键信息抽取-SER+RE服务包
deploy/hubserving/structure_system/└─__init__.py空文件,必选└─config.json配置文件,可选,使用配置启动服务时作为参数传入└─module.py主模块,必选,包含服务的完整逻辑└─params.py参数文件,必选,包含模型路径、前后处理参数等参数
| 模型 | 路径 |
| 检测模型 | ./inference/ch_PP-OCRv4_det_infer/ |
| 识别模型 | ./inference/ch_PP-OCRv4_rec_infer/ |
| 方向分类器 | ./inference/ch_ppocr_mobile_v2.0_cls_infer/ |
| 表格识别模型 | ./inference/ch_ppstructure_mobile_v2.0_SLANet_infer/ |
| 版面分析模型 | ./inference/picodet_lcnet_x1_0_fgd_layout_cdla_infer/ |
模型路径可在params.py中查看和修改,可以替换成自己训练转换好的模型。
/替换为\)下,安装模块命令如下表:| 服务模块 | 命令 |
| 全服务串联 | hub install deploy/hubserving/structure_system |
#执行全服务模型安装命令hub install deploy/hubserving/structure_system#执行后可以看到以下输出,说名全服务串联模型安装成功[2024-10-10 21:56:43,853] [INFO] - Successfully uninstalled structure_system[2024-10-1021:56:44,570][INFO]-Successfullyinstalledstructure_system-1.0.0
3. 启动服务
hubservingstart--modulesModule1==Version1,Module2==Version2,...\--port8866\--use_multiprocess\--workers\
--modules/-m | PaddleHub Serving预安装模型,以多个Module==Version键值对的形式列出 当不指定Version时,默认选择最新版本 |
--port/-p | 服务端口,默认为8866 |
| --use_multiprocess | 是否启用并发方式,默认为单进程方式,推荐多核CPU机器使用此方式 Windows操作系统只支持单进程方式 |
| --workers | 在并发方式下指定的并发任务数,默认为2*cpu_count-1,其中cpu_count为CPU核数 |
#执行后会出现以下图片输出,说明服务已经启动hubservingstart-mstructure_system
4. 请求服务
进入E:\paddle_ocr\PaddleOCR源码路径下,在praddle_env的Python的环境下,执行以下命令,验证已启动服务是否可用。
python./tools\test_hubserving.py--server_url=http://127.0.0.1:8866/predict/structure_system--image_dir=./doc\imgs/train_1430.jpg--visualize=true
{'regions':[{'bbox':[178,161,769,214],'img_idx':0,'res':[{'confidence':0.9866982698440552,'text':'(二)现任及报告期内离任董事、监事和高级管理人员的任职情况','text_region':[[179.0,160.0],[773.0,160.0],[773.0,188.0],[179.0,188.0]]},{'confidence':0.996711254119873,'text':'1.','text_region':[[179.0,195.0],[223.0,195.0],[223.0,214.0],[179.0,214.0]]},{'confidence':0.9996092319488525,'text':'在股东单位任职情况','text_region':[[213.0,191.0],[415.0,191.0],[415.0,219.0],[213.0,219.0]]}],'score':0.8759905099868774,'type':'text'},{'bbox':[160,563,1083,1523],'img_idx':0,'res':{'cell_bbox':[[4.498061656951904,9.545287132263184,123.835205078125,9.642200469970703,121.98422241210938,56.403995513916016,4.303066730499268,56.22964859008789],[125.66343688964844,10.86853313446045,601.1064453125,10.622770309448242,600.7944946289062,44.1065788269043,124.36930847167969,45.06888961791992],[503.7535095214844,7.360479354858398,701.1563110351562,7.2659502029418945,702.7037353515625,36.61697769165039,503.8215637207031,37.77928924560547],[670.3419799804688,8.419022560119629,811.7805786132812,8.446755409240723,812.40380859375,34.07345199584961,671.380126953125,34.663631439208984],[784.5574340820312,7.437520503997803,916.24853515625,7.580190658569336,916.2015380859375,37.95753479003906,784.3206787109375,38.26716232299805],[12.037315368652344,45.964595794677734,111.65966033935547,44.94559860229492,112.05738830566406,75.57157897949219,11.772783279418945,76.49816131591797],[118.06964111328125,44.1462516784668,523.6019897460938,43.5856819152832,520.7415771484375,76.78903198242188,116.08219146728516,77.10972595214844],[513.0073852539062,34.979522705078125,695.7911987304688,34.823394775390625,694.02978515625,75.75557708740234,510.40142822265625,76.450439453125],[684.157958984375,30.152864456176758,813.8160400390625,30.23993682861328,812.6011352539062,74.0234603881836,683.2059326171875,74.5884780883789],[798.927978515625,25.941539764404297,920.3408203125,26.354337692260742,920.264892578125,80.3907699584961,797.3761596679688,80.34146118164062],[7.0165510177612305,75.14842224121094,114.26177978515625,74.92013549804688,114.26763916015625,104.26825714111328,6.945335865020752,104.322998046875],[102.17969512939453,77.53343200683594,518.8002319335938,77.28779602050781,515.6495971679688,103.47911071777344,100.58755493164062,103.41241455078125],[507.00738525390625,74.96455383300781,675.162109375,74.89897918701172,672.6912231445312,102.4069595336914,504.8054504394531,102.30765533447266],[678.404052734375,73.607421875,800.37060546875,73.51984405517578,798.9258422851562,100.8022689819336,676.9501342773438,100.92758178710938],[796.0067749023438,72.87516784667969,915.516845703125,73.34247589111328,915.386962890625,103.08671569824219,795.2428588867188,102.49726867675781],[6.861465930938721,102.2308120727539,116.91632843017578,102.12569427490234,117.12047576904297,130.55906677246094,6.818292140960693,130.92625427246094],[110.77519989013672,103.46292877197266,524.7883911132812,103.16129302978516,522.3780517578125,129.6745147705078,109.1706314086914,129.89244079589844],[507.64862060546875,101.51544189453125,679.7174682617188,101.47674560546875,676.9719848632812,127.806640625,504.9059753417969,127.89181518554688],[678.2163696289062,100.89976501464844,799.0449829101562,100.82760620117188,797.560791015625,126.18241119384766,676.17626953125,126.42273712158203],[795.289306640625,100.16848754882812,917.5969848632812,100.89277648925781,917.4988403320312,130.31466674804688,794.5338745117188,129.49441528320312],[6.309231281280518,129.81033325195312,117.75189971923828,129.7296142578125,118.10103607177734,158.37420654296875,6.271739482879639,158.72708129882812],[111.74261474609375,130.5666046142578,523.5564575195312,130.32449340820312,521.0000610351562,157.00025939941406,110.11062622070312,157.1968231201172],[509.80743408203125,128.63839721679688,679.2877807617188,128.78741455078125,676.416259765625,155.45098876953125,506.5506591796875,155.3666229248047],[680.0029907226562,128.4838409423828,798.1373291015625,128.5179901123047,796.6524658203125,154.404052734375,678.0977783203125,154.4539031982422],[796.6156616210938,128.10235595703125,916.9672241210938,129.07981872558594,916.8707275390625,158.44549560546875,796.0604858398438,157.3574676513672],[7.2437825202941895,158.40318298339844,120.37187957763672,158.58970642089844,120.48767852783203,185.87872314453125,7.170042037963867,186.1480712890625],[113.90608978271484,158.3458709716797,519.2463989257812,158.2955322265625,516.5323486328125,184.3142852783203,112.19956970214844,184.4180908203125],[512.0233764648438,157.19541931152344,678.2108764648438,157.4663543701172,675.2616577148438,183.47076416015625,508.43017578125,183.39646911621094],[681.6056518554688,157.08363342285156,796.3106689453125,157.13516235351562,794.8079833984375,182.3052215576172,679.674560546875,182.4419403076172],[796.5963134765625,156.60037231445312,917.1970825195312,157.83358764648438,917.1041870117188,186.4396514892578,796.000244140625,185.2259979248047],[6.7951531410217285,186.40965270996094,117.11863708496094,187.03765869140625,117.21306610107422,213.91928100585938,6.704384803771973,213.70359802246094],[114.6916732788086,185.79824829101562,515.8260498046875,186.38291931152344,512.8416748046875,211.52391052246094,112.82210540771484,211.0830078125],[515.2515258789062,185.31101989746094,688.7913208007812,185.75274658203125,685.8671264648438,211.8577880859375,511.336181640625,211.62953186035156],[680.2713623046875,185.1221466064453,795.8648071289062,185.34738159179688,794.3031616210938,210.09091186523438,678.247314453125,210.11669921875],[794.1371459960938,184.33921813964844,917.852294921875,186.01409912109375,917.7756958007812,213.7703857421875,793.6712646484375,212.2020263671875],[6.29174280166626,213.36239624023438,113.37252807617188,214.33297729492188,113.53797912597656,242.56422424316406,6.203976154327393,241.97682189941406],[110.78193664550781,213.397705078125,513.780029296875,214.36888122558594,510.796875,240.97987365722656,108.94802856445312,240.23928833007812],[515.1849975585938,212.7093505859375,680.4038696289062,213.57620239257812,677.36083984375,241.09719848632812,511.23760986328125,240.4213104248047],[679.7124633789062,212.70578002929688,796.93603515625,213.2115478515625,795.3026123046875,238.86875915527344,677.57373046875,238.54681396484375],[793.6317749023438,212.3407440185547,917.0255737304688,214.4727783203125,916.9485473632812,242.10833740234375,793.383544921875,240.02752685546875],[6.5448198318481445,241.24432373046875,114.20539093017578,242.67388916015625,114.13536071777344,271.2123718261719,6.439479351043701,270.351806640625],[116.98934936523438,241.564697265625,517.6928100585938,242.8634033203125,514.4597778320312,268.9460754394531,114.92552947998047,268.0157470703125],[516.0845336914062,241.01336669921875,679.1426391601562,241.9478759765625,676.0394287109375,269.112060546875,512.0684814453125,268.5206298828125],[679.0955810546875,240.81900024414062,796.73193359375,241.51107788085938,794.986328125,266.8581237792969,676.7041015625,266.54541015625],[792.6276245117188,240.69056701660156,917.088134765625,243.08409118652344,917.0087280273438,271.10858154296875,792.287353515625,268.9478759765625],[5.400808811187744,268.5782775878906,112.3357925415039,270.30792236328125,112.48979187011719,298.1329650878906,5.315079212188721,296.958251953125],[112.02423858642578,269.35028076171875,518.093994140625,270.7965087890625,515.220458984375,297.40966796875,110.16490936279297,296.407958984375],[517.510498046875,268.68438720703125,676.3302001953125,270.0392150878906,673.237060546875,298.1700439453125,513.5993041992188,297.159423828125],[679.753662109375,268.6905212402344,795.8819580078125,269.49578857421875,794.1557006835938,295.546630859375,677.4940795898438,294.9688720703125],[793.2979125976562,268.1053466796875,916.9616088867188,270.66961669921875,916.8848266601562,298.6878662109375,793.1030883789062,296.2706604003906],[5.867050647735596,296.66815185546875,112.93968963623047,298.5991516113281,112.9771957397461,324.9601745605469,5.779616832733154,323.63623046875],[115.33856964111328,298.0686950683594,516.7718505859375,299.5196228027344,514.2528076171875,325.25872802734375,113.65840911865234,324.24444580078125],[518.3207397460938,297.8866882324219,678.3632202148438,299.10650634765625,675.495361328125,325.8212890625,514.7550659179688,324.986328125],[680.27783203125,297.2555236816406,796.1353149414062,298.005615234375,794.4814453125,322.715087890625,678.1358032226562,322.2394714355469],[793.6611328125,296.85858154296875,917.1815795898438,299.3363037109375,917.1093139648438,326.69586181640625,793.494140625,324.3544616699219],[6.1058454513549805,324.3236999511719,112.74211120605469,326.30908203125,112.72953033447266,352.72625732421875,6.020620346069336,351.2809753417969],[115.44058990478516,325.4470520019531,513.4751586914062,327.02362060546875,511.01556396484375,353.0732421875,113.80284881591797,351.9068603515625],[517.5945434570312,325.1158752441406,683.7572631835938,326.23883056640625,680.767578125,353.0166931152344,513.843017578125,352.23297119140625],[679.9055786132812,324.68585205078125,795.2625732421875,325.2931823730469,793.519775390625,350.2931213378906,677.6004638671875,349.8802490234375],[793.8165283203125,324.39276123046875,917.3563842773438,326.90093994140625,917.2875366210938,353.9791564941406,793.7106323242188,351.5252990722656],[6.120126724243164,352.3223876953125,113.45146942138672,354.4226989746094,113.4281997680664,381.52862548828125,6.0364885330200195,379.93212890625],[116.53500366210938,353.32720947265625,513.62158203125,354.8731994628906,511.34027099609375,381.77099609375,114.94812774658203,380.5647277832031],[517.422607421875,352.5205078125,684.0896606445312,353.67352294921875,681.1389770507812,380.9615173339844,513.6475219726562,380.1260070800781],[680.4839477539062,352.1560363769531,795.5714721679688,352.74578857421875,793.840087890625,378.3210754394531,678.1736450195312,377.876708984375],[795.0482177734375,351.76434326171875,917.20166015625,354.3714294433594,917.1363525390625,381.8531494140625,795.0171508789062,379.2604064941406],[6.023154258728027,380.89715576171875,115.20487213134766,383.19482421875,115.1811752319336,410.4507751464844,5.940659999847412,408.65252685546875],[117.08432006835938,381.9545593261719,515.3921508789062,383.5587158203125,512.9667358398438,410.3992919921875,115.40318298339844,409.08282470703125],[519.7257080078125,381.23907470703125,681.6235961914062,382.3709716796875,678.6065063476562,409.12738037109375,515.8543090820312,408.27325439453125],[680.9976806640625,380.9011535644531,794.9946899414062,381.4741516113281,793.2182006835938,407.1292419433594,678.6024780273438,406.6338806152344],[794.4586791992188,380.3659362792969,917.2334594726562,383.09210205078125,917.1670532226562,411.13885498046875,794.3790283203125,408.412841796875],[5.892773151397705,409.499267578125,117.3931884765625,412.0484619140625,117.45379638671875,439.5044860839844,5.811656475067139,437.40777587890625],[116.28411102294922,411.0489501953125,514.0737915039062,412.8046569824219,511.8162536621094,440.2318115234375,114.64078521728516,438.7586364746094],[520.4713745117188,409.98419189453125,688.3170776367188,411.1687316894531,685.4898681640625,438.53271484375,516.751953125,437.54608154296875],[681.65673828125,409.6797790527344,794.018310546875,410.3707580566406,792.228759765625,436.8336486816406,679.20166015625,436.1838684082031],[795.27294921875,408.9578857421875,916.5808715820312,411.9627380371094,916.5090942382812,440.3974304199219,795.1828002929688,437.3348388671875],[6.006280422210693,438.017578125,117.17462921142578,440.698974609375,117.12943267822266,468.03082275390625,5.917118072509766,465.7499084472656],[118.75875854492188,439.47552490234375,514.1254272460938,441.36407470703125,511.8039855957031,467.88092041015625,117.05877685546875,466.27252197265625],[520.7886352539062,438.5564880371094,689.74365234375,439.66766357421875,687.0170288085938,466.8045349121094,517.2111206054688,465.8734130859375],[681.4100341796875,438.1256408691406,792.4796752929688,438.8173522949219,790.7383422851562,465.271240234375,679.0089721679688,464.611083984375],[794.5274658203125,437.79534912109375,916.4288940429688,440.8071594238281,916.3577270507812,469.1437072753906,794.4360961914062,466.1128234863281],[6.646564960479736,465.1865234375,116.77031707763672,467.8735046386719,116.62709045410156,495.6376647949219,6.547154426574707,493.3059387207031],[114.22871398925781,466.6871337890625,513.6736450195312,468.6549987792969,511.52020263671875,496.5535888671875,112.63668060302734,494.79144287109375],[521.4395751953125,465.9921875,685.6351928710938,467.25933837890625,682.9611206054688,495.03009033203125,518.0394287109375,493.82733154296875],[681.3358154296875,465.76568603515625,792.577880859375,466.5036926269531,790.916015625,493.72735595703125,679.0562744140625,492.8785400390625],[794.7592163085938,465.6238098144531,916.0322875976562,468.62969970703125,915.96435546875,497.22015380859375,794.7368774414062,494.10028076171875],[7.080552577972412,492.99871826171875,115.35064697265625,495.6551818847656,115.0821533203125,523.73583984375,6.970582008361816,521.4666748046875],[117.12582397460938,495.0509338378906,513.2804565429688,496.9726257324219,511.1614074707031,524.52783203125,115.54479217529297,522.7581176757812],[520.9531860351562,494.3744201660156,686.89501953125,495.5498352050781,684.3833618164062,523.1033325195312,517.73828125,521.9557495117188],[681.6245727539062,494.384033203125,795.3270263671875,495.10064697265625,793.76220703125,522.2477416992188,679.509765625,521.4126586914062],[795.0294799804688,494.21539306640625,916.3636474609375,497.1510925292969,916.3019409179688,526.214111328125,795.0560302734375,523.1083374023438],[6.4729108810424805,519.9852905273438,116.83475494384766,522.5755615234375,116.60794067382812,551.71337890625,6.370972633361816,549.3638305664062],[115.9369125366211,521.8094482421875,516.6080932617188,523.8027954101562,514.4926147460938,552.9054565429688,114.35472869873047,551.0243530273438],[521.470703125,521.4610595703125,685.2645263671875,522.7974853515625,682.819580078125,552.2919921875,518.4140625,550.852294921875],[682.4373168945312,522.379150390625,795.1343994140625,523.1663818359375,793.5913696289062,551.9328002929688,680.4415893554688,550.8289184570312],[794.2073974609375,521.7266235351562,917.01806640625,524.7412109375,916.967529296875,555.4213256835938,794.4218139648438,552.0775756835938],[6.730972766876221,547.187744140625,118.5295639038086,550.00830078125,118.16275024414062,579.5826416015625,6.6192121505737305,577.058837890625],[118.13018035888672,550.26953125,517.9166259765625,552.295654296875,515.9754028320312,581.7735595703125,116.59608459472656,579.7977905273438],[522.2399291992188,550.0038452148438,684.2053833007812,551.223876953125,681.8827514648438,580.8683471679688,519.3905029296875,579.5255737304688],[683.2606201171875,550.7817993164062,794.67822265625,551.5228881835938,793.15380859375,580.715576171875,681.33056640625,579.62939453125],[793.6685180664062,549.760009765625,917.4904174804688,552.6762084960938,917.44921875,584.2667846679688,794.003662109375,580.9561767578125],[6.077193260192871,574.877685546875,121.26602935791016,577.7805786132812,120.88319396972656,607.422119140625,5.976038455963135,604.8021240234375],[118.7756576538086,578.5208740234375,523.3243408203125,580.6303100585938,521.5650024414062,609.4354858398438,117.23011016845703,607.4308471679688],[521.6472778320312,578.6497192382812,684.4058837890625,579.8702392578125,682.12158203125,608.9528198242188,518.9239501953125,607.59765625],[681.9610595703125,579.0086669921875,793.6410522460938,579.630615234375,792.0474243164062,608.5929565429688,680.0015258789062,607.5089721679688],[793.8855590820312,578.4741821289062,917.5005493164062,581.1654052734375,917.4595336914062,612.7914428710938,794.2481079101562,609.6333618164062],[6.378926753997803,601.6633911132812,120.84318542480469,604.6452026367188,120.43370819091797,634.3651123046875,6.278337001800537,631.770263671875],[118.65218353271484,606.8605346679688,518.2590942382812,608.838134765625,516.8697509765625,637.6775512695312,117.22348022460938,635.790283203125],[519.7401733398438,607.0062866210938,684.102294921875,608.1957397460938,681.9610595703125,636.6435546875,517.17529296875,635.3145141601562],[682.439208984375,607.2513427734375,794.5999755859375,607.7946166992188,793.0851440429688,636.4054565429688,680.5725708007812,635.4327392578125],[794.9539184570312,606.6293334960938,917.4075927734375,609.0800170898438,917.36865234375,640.3255615234375,795.319091796875,637.3814086914062],[6.260758399963379,629.4912109375,118.84089660644531,632.2918090820312,118.41433715820312,661.6791381835938,6.158733367919922,659.261962890625],[119.88860321044922,634.0977783203125,515.8519287109375,635.89501953125,514.4122314453125,664.1223754882812,118.41728210449219,662.4235229492188],[519.30419921875,634.2418212890625,682.0534057617188,635.1372680664062,679.9108276367188,663.2901000976562,516.7450561523438,662.257568359375],[682.47216796875,634.9707641601562,794.562744140625,635.3099365234375,793.0394287109375,663.774658203125,680.5489501953125,663.0133056640625],[795.1912841796875,634.296142578125,917.4498291015625,636.51123046875,917.4081420898438,668.0294189453125,795.4437866210938,665.3713989257812],[6.2009148597717285,657.3807373046875,123.59737396240234,660.0474853515625,123.1484375,688.7388305664062,6.098379611968994,686.3450927734375],[114.9033432006836,660.3289184570312,520.0084838867188,661.9470825195312,518.6936645507812,690.9708251953125,113.45858001708984,689.385498046875],[519.4085693359375,661.093017578125,674.1309204101562,661.8583984375,671.9353637695312,690.7783813476562,516.8613891601562,689.75634765625],[682.7269897460938,662.0474853515625,793.76123046875,662.3713989257812,792.2555541992188,691.1878662109375,680.8466796875,690.370361328125],[795.9068603515625,661.1119384765625,916.5740966796875,663.2567138671875,916.5292358398438,694.6463012695312,796.1234130859375,692.0155029296875],[6.2657599449157715,685.8135986328125,126.13443756103516,688.550048828125,125.43482208251953,716.8471069335938,6.142832279205322,714.4879760742188],[119.63490295410156,689.0230102539062,522.14404296875,690.4298706054688,520.9368286132812,719.0474243164062,118.10336303710938,717.6790161132812],[520.7330322265625,688.894287109375,677.448486328125,689.318115234375,675.3591918945312,717.6685180664062,518.2669067382812,717.1019897460938],[683.5082397460938,689.4368286132812,794.3580322265625,689.6461181640625,792.9151000976562,718.3873291015625,681.5901489257812,717.7964477539062],[795.6759643554688,688.61376953125,916.5652465820312,690.58154296875,916.5197143554688,722.379150390625,795.8627319335938,720.0523681640625],[5.323832988739014,714.85302734375,127.40572357177734,717.3871459960938,126.57168579101562,746.0682373046875,5.204966068267822,743.8416137695312],[115.09811401367188,716.8601684570312,522.6958618164062,718.2274780273438,521.2353515625,747.5923461914062,113.41251373291016,746.3402099609375],[520.5660400390625,716.7273559570312,674.061767578125,717.2669677734375,671.7724609375,745.3856201171875,517.8644409179688,744.73046875],[682.341552734375,717.3255615234375,794.8858642578125,717.6165161132812,793.3638305664062,746.0123901367188,680.252197265625,745.3659057617188],[795.5003051757812,716.3673706054688,916.1456909179688,718.3086547851562,916.1008911132812,749.394775390625,795.7312622070312,747.0814208984375],[5.440361976623535,743.7098388671875,127.05612182617188,745.9974975585938,126.19889068603516,774.5255126953125,5.3206963539123535,772.4663696289062],[115.65984344482422,746.0537719726562,517.32080078125,747.1287231445312,515.9242553710938,776.246826171875,114.01081848144531,775.2319946289062],[518.5719604492188,746.4093627929688,685.2074584960938,746.8080444335938,683.1649780273438,774.546875,515.9572143554688,773.9979248046875],[682.1038818359375,746.4907836914062,796.7066650390625,746.5672607421875,795.37890625,774.288818359375,680.3252563476562,773.8843383789062],[795.0401000976562,745.7349243164062,916.5875244140625,747.3155517578125,916.5481567382812,777.6805419921875,795.3184814453125,775.7805786132812],[5.333799362182617,772.6932983398438,119.861083984375,774.3359985351562,119.14483642578125,803.3704833984375,5.224424839019775,801.8425903320312],[112.84999084472656,774.7462158203125,519.9200439453125,775.3406982421875,518.67236328125,805.279296875,111.31723022460938,804.6646728515625],[518.6363525390625,774.6973266601562,683.896484375,774.9473266601562,681.99072265625,803.3995361328125,516.2350463867188,802.9638671875],[683.1265258789062,774.5504760742188,797.0020141601562,774.3455200195312,795.7554931640625,802.0222778320312,681.59716796875,801.8220825195312],[794.1913452148438,774.3096313476562,917.1387329101562,775.4821166992188,917.1097412109375,805.446044921875,794.697509765625,803.9927368164062],[5.889903545379639,800.8614501953125,118.84339904785156,802.2684936523438,118.09706115722656,831.2734375,5.778231620788574,829.9820556640625],[117.87205505371094,802.9805908203125,523.6100463867188,803.5966186523438,522.6058959960938,832.6526489257812,116.36421203613281,832.0724487304688],[520.0425415039062,802.065673828125,675.6690673828125,802.343505859375,673.74560546875,830.6536254882812,517.6033935546875,830.3307495117188],[684.1702880859375,801.8265380859375,797.237060546875,801.6825561523438,796.1069946289062,829.626708984375,682.7803955078125,829.4955444335938],[793.947998046875,801.77783203125,917.4400634765625,802.7328491210938,917.4224853515625,832.6280517578125,794.6630249023438,831.4973754882812],[4.455350399017334,830.5655517578125,122.71109771728516,831.690185546875,122.52877807617188,862.5700073242188,4.398390769958496,861.603515625],[129.7024688720703,833.6222534179688,525.7745971679688,834.4804077148438,525.2967529296875,861.88037109375,128.3724365234375,861.2626342773438],[520.3342895507812,832.277099609375,676.1993408203125,832.5933837890625,674.6917724609375,860.7274169921875,518.3392944335938,860.5776977539062],[685.20947265625,831.5123901367188,797.5484619140625,831.35595703125,796.6613159179688,860.2808837890625,684.2706909179688,860.2941284179688],[794.7259521484375,831.8392333984375,918.1787109375,832.3438720703125,918.1824951171875,862.59765625,795.9983520507812,861.9784545898438],[4.116330623626709,860.6699829101562,122.2989501953125,861.576171875,122.50164031982422,893.0823364257812,4.095632076263428,892.4369506835938],[126.02391052246094,863.6384887695312,514.0679321289062,864.6732788085938,514.460693359375,892.21240234375,125.21572875976562,891.6570434570312],[520.2794189453125,862.1533203125,678.2034301757812,862.4281005859375,676.9805908203125,890.91748046875,518.5653076171875,890.8864135742188],[685.7990112304688,861.2681274414062,797.51953125,861.0469360351562,796.8839111328125,890.5406494140625,685.2905883789062,890.7236328125],[794.6419067382812,862.0407104492188,918.473388671875,862.1033935546875,918.4985961914062,891.7745971679688,796.538330078125,891.6215209960938],[5.611627578735352,889.2611694335938,118.41958618164062,890.1727905273438,118.49215698242188,920.4960327148438,5.58912992477417,920.0267333984375],[126.48420715332031,891.0299682617188,505.885498046875,892.060791015625,506.22467041015625,919.0881958007812,125.57593536376953,918.6825561523438],[518.335693359375,890.2766723632812,677.5947265625,890.6682739257812,676.1796264648438,918.2865600585938,515.9285888671875,918.2697143554688],[683.6793823242188,890.0584716796875,797.1397094726562,890.017578125,796.5440063476562,918.36865234375,682.8662109375,918.5692138671875],[793.0550537109375,890.6959228515625,917.86328125,890.7040405273438,917.8871459960938,919.0569458007812,794.7377319335938,919.1201171875],[4.62816858291626,920.2601318359375,119.03997802734375,921.25,118.21710968017578,950.4440307617188,4.560724258422852,950.134765625],[120.43228912353516,922.2282104492188,500.5233154296875,923.2161254882812,500.7606506347656,949.262939453125,118.76371002197266,948.9961547851562],[513.6383666992188,922.5336303710938,665.0474243164062,923.1267700195312,663.9823608398438,948.9828491210938,510.3147277832031,948.8202514648438],[684.0032958984375,922.842529296875,796.522705078125,923.1132202148438,796.1115112304688,948.7281494140625,682.1686401367188,948.6721801757812],[794.3134765625,922.5149536132812,916.2750244140625,922.6748657226562,916.269287109375,947.96728515625,794.5840454101562,947.9280395507812]],'html':'<html><body><table><tbody><tr><td>√适用口不适用任职人员</td><td></td><td>在其他单位担任</td><td>任期起始</td><td>任期终止</td></tr><tr><td>姓名</td><td>其他单位名称</td><td>的职务</td><td>日期</td><td>日期</td></tr><tr><td>袁义祥</td><td>大连和升控股集团有限公司</td><td>董事</td><td>2008-04</td><td></td></tr><tr><td>袁义祥</td><td>大连三鑫投资有限公司</td><td>监事</td><td>2014-09</td><td></td></tr><tr><td>袁义祥</td><td>大连装备创新与投资有限公司</td><td>监事</td><td>2015-12</td><td></td></tr><tr><td>袁义祥</td><td>大连百傲化学股份有限公司</td><td>董事</td><td>2015-06</td><td></td></tr><tr><td>袁义祥</td><td>沈阳和怡新材料有限公司</td><td>执行董事、经理</td><td>2020-07</td><td></td></tr><tr><td>袁义祥</td><td>和升实业有限公司</td><td>董事</td><td>2017-06</td><td></td></tr><tr><td>袁义祥</td><td>润新微电子(大连)有限公司</td><td>监事会主席</td><td>2022-05</td><td></td></tr><tr><td>袁义祥</td><td>辽宁省机械(集团)股份有限公司</td><td>董事</td><td>2012-08</td><td></td></tr><tr><td>袁义祥</td><td>大连汇宇鑫科技有限公司</td><td>董事</td><td>2006-07</td><td></td></tr><tr><td>袁义祥</td><td>大连德润电动汽车科技有限公司</td><td>副董事长</td><td>2017-03</td><td></td></tr><tr><td>袁义祥</td><td>爱康普科技(大连)有限公司</td><td>董事</td><td>2016-04</td><td></td></tr><tr><td>袁义祥</td><td>大连业通投资有限公司</td><td>监事</td><td>2019-04</td><td></td></tr><tr><td>袁义祥</td><td>大连亿阳信息技术有限公司</td><td>执行董事、经理</td><td>2022-05</td><td></td></tr><tr><td>王文锋</td><td>大连和升控股集团有限公司</td><td>董事长、总经理</td><td>2007-07</td><td></td></tr><tr><td>王文锋</td><td>大连辽机路航特种车制造有限公司</td><td>董事长</td><td>2010-01</td><td></td></tr><tr><td>王文锋</td><td>大连三鑫投资有限公司</td><td>执行董事、经理</td><td>2010-01</td><td></td></tr><tr><td>王文锋</td><td>大连百傲化学股份有限公司</td><td>副董事长</td><td>2015-06</td><td></td></tr><tr><td>王文锋</td><td>和升实业有限公司</td><td>董事长、经理</td><td>2017-06</td><td></td></tr><tr><td>王文锋</td><td>北京和升创展食品发展有限责任公司</td><td>董事长、经理</td><td>2019-05</td><td></td></tr><tr><td>王文锋</td><td>大连听心会议服务有限公司</td><td>执行董事、经理</td><td>2020-09</td><td></td></tr><tr><td>王文锋</td><td>大连汇宇鑫科技有限公司</td><td>董事</td><td>2017-07</td><td></td></tr><tr><td>韩东丰</td><td>大连和升控股集团有限公司</td><td>董事、经理</td><td>2018-03</td><td></td></tr><tr><td>韩东丰</td><td>大连汇宇鑫科技有限公司</td><td>董事</td><td>2017-07</td><td></td></tr><tr><td>韩东丰</td><td>南京兰埔成新材料有限公司</td><td>董事</td><td>2016-04</td><td></td></tr><tr><td>韩东丰</td><td>大连万怡投资有限公司</td><td>执行董事、经理</td><td>2014-10</td><td></td></tr><tr><td>韩东丰</td><td>新大洲控股股份有限公司</td><td>董事长</td><td>2020-09-18</td><td></td></tr><tr><td>韩东丰</td><td>北京和升创展食品发展有限责任公司</td><td>董事</td><td>2019-05</td><td></td></tr><tr><td>韩东丰</td><td>沈阳和怡新材料有限公司</td><td>董事</td><td>2021-09</td><td></td></tr><tr><td>韩东丰</td><td>大连四达高技术发展有限公司</td><td>董事</td><td>2019-07</td><td></td></tr><tr><td>韩东丰</td><td>大连联创智达信息技术有限公司</td><td>董事</td><td>2019-10</td><td></td></tr><tr><td>韩东丰</td><td>盛银消费金融有限公司</td><td>董事</td><td>2019-08</td><td></td></tr></tbody></table></body></html>'},'score':0.977480947971344,'type':'table'},{'bbox':[161,225,1085,507],'img_idx':0,'res':{'cell_bbox':[[5.785220623016357,24.311243057250977,181.07508850097656,23.637800216674805,185.1851348876953,72.7176742553711,5.823643684387207,73.16725158691406],[178.80596923828125,22.35544204711914,419.8334655761719,21.064998626708984,425.7052307128906,63.92222213745117,180.6250762939453,66.30516815185547],[400.674072265625,9.518608093261719,606.553466796875,9.219185829162598,610.5952758789062,69.24838256835938,403.7786560058594,72.22607421875],[581.6190795898438,18.830272674560547,768.76611328125,18.454299926757812,769.8507690429688,67.78468322753906,584.0712890625,69.25111389160156],[748.94189453125,19.57159423828125,914.4827880859375,19.501468658447266,914.4170532226562,72.4549560546875,750.02880859375,72.8739242553711],[6.469827651977539,67.82855224609375,165.38909912109375,67.83541107177734,165.43399047851562,103.52085876464844,6.291061878204346,103.28355407714844],[173.060546875,67.76287078857422,422.35882568359375,67.77702331542969,421.85748291015625,99.12926483154297,170.94956970214844,99.13125610351562],[416.9671936035156,62.83456039428711,587.1475830078125,63.146575927734375,585.9091796875,100.19012451171875,414.5832824707031,99.99385833740234],[592.1345825195312,64.03622436523438,750.5368041992188,63.95559310913086,749.17626953125,97.11198425292969,589.9861450195312,97.40837860107422],[780.7509155273438,63.151573181152344,911.5607299804688,63.30153274536133,911.3961181640625,97.13643646240234,779.7352294921875,97.09942626953125],[6.2628092765808105,99.06214904785156,170.69480895996094,99.63970947265625,170.6552276611328,129.71847534179688,6.143228054046631,129.2323760986328],[173.407958984375,98.6748046875,420.6581115722656,98.76355743408203,419.62921142578125,127.04004669189453,171.16477966308594,127.03429412841797],[416.32757568359375,96.16886901855469,584.0556640625,96.40225219726562,582.3076171875,124.58045959472656,413.4111633300781,124.51908874511719],[590.5037231445312,96.67974090576172,752.9931030273438,96.6097640991211,751.8983764648438,124.3765640258789,588.7061157226562,124.61097717285156],[785.4635620117188,96.46088409423828,912.8157348632812,96.69051361083984,912.662841796875,126.30024719238281,784.5137329101562,126.12730407714844],[5.720996856689453,126.72210693359375,172.47499084472656,127.42425537109375,172.48741149902344,158.52371215820312,5.623476982116699,157.99790954589844],[174.05043029785156,126.45726013183594,418.9090576171875,127.03121185302734,417.4536437988281,156.33580017089844,171.69680786132812,155.88070678710938],[418.59783935546875,124.23936462402344,587.9127197265625,124.83297729492188,585.6192016601562,154.0175323486328,414.9079284667969,153.5950469970703],[593.3422241210938,124.64849090576172,757.4677734375,124.85608673095703,756.1747436523438,153.73577880859375,591.0472412109375,153.64122009277344],[786.3098754882812,124.77445983886719,913.7241821289062,125.28885650634766,913.5853881835938,155.95135498046875,785.4411010742188,155.41098022460938],[5.801194667816162,155.31520080566406,176.6939239501953,156.4505157470703,176.1646270751953,187.27545166015625,5.674062252044678,186.305419921875],[172.2061767578125,154.86753845214844,419.05615234375,155.85899353027344,417.4673156738281,185.4031219482422,169.7149658203125,184.5210418701172],[414.32965087890625,152.75909423828125,592.2716674804688,153.6141357421875,590.1124267578125,182.86080932617188,410.3360290527344,182.1475372314453],[589.1049194335938,152.78106689453125,762.3895874023438,153.16416931152344,761.2607421875,182.85633850097656,586.8853759765625,182.54150390625],[786.94384765625,153.03585815429688,913.7927856445312,153.81143188476562,913.6409301757812,185.2374725341797,785.9165649414062,184.50112915039062],[6.30924129486084,183.93406677246094,173.02272033691406,185.31723022460938,171.75796508789062,219.1966094970703,6.0957136154174805,218.19049072265625],[176.25685119628906,182.89517211914062,418.9935607910156,184.02035522460938,416.5292663574219,214.67453002929688,172.76731872558594,213.8861541748047],[416.76849365234375,180.90020751953125,594.4534912109375,181.86082458496094,592.14599609375,212.08982849121094,411.8827819824219,211.4750213623047],[591.6781005859375,181.07540893554688,765.05712890625,181.67526245117188,763.4891967773438,211.99407958984375,588.595947265625,211.65943908691406],[791.48828125,181.5181121826172,914.7260131835938,182.6168212890625,914.5346069335938,215.15953063964844,789.9700317382812,214.34336853027344],[4.574718475341797,215.89959716796875,174.4020538330078,216.35736083984375,170.20762634277344,276.5448303222656,4.326537609100342,276.3892822265625],[166.39047241210938,215.0373992919922,745.2926025390625,215.20089721679688,742.1246948242188,268.39129638671875,162.07473754882812,268.2206115722656],[753.2962646484375,215.2685089111328,906.15283203125,216.4825897216797,906.400634765625,268.064453125,753.3102416992188,267.6579895019531]],'html':'<html><body><table><tbody><tr><td>√适用口不适用任职人员姓名</td><td>股东单位名称</td><td>在股东单位担任的职务</td><td>任期起始日期</td><td>任期终止日期</td></tr><tr><td>袁义祥</td><td>亿阳集团股份有限公司</td><td>董事</td><td>2021-01</td><td></td></tr><tr><td>王文锋</td><td>亿阳集团股份有限公司</td><td>董事</td><td>2021-01</td><td></td></tr><tr><td>韩东丰</td><td>亿阳集团股份有限公司</td><td>监事长</td><td>2021-01</td><td></td></tr><tr><td>韩东丰</td><td>大连万怡投资有限公司</td><td>法人、执行董事</td><td>2014-10</td><td></td></tr><tr><td>张秀琴</td><td>大连万怡投资有限公司</td><td>监事</td><td>2014-10</td><td></td></tr><tr><td>在股东单位任职情况的说明</td><tdcolspan="4">无</td><td></td></tr></tbody></table></body></html>'},'score':0.9516164064407349,'type':'table'},{'bbox':[180,530,399,549],'img_idx':0,'res':[{'confidence':0.9975435137748718,'text':'2.','text_region':[[177.0,528.0],[225.0,528.0],[225.0,553.0],[177.0,553.0]]},{'confidence':0.9998181462287903,'text':'在其他单位任职情况','text_region':[[213.0,526.0],[415.0,526.0],[415.0,556.0],[213.0,556.0]]}],'score':0.8662838935852051,'type':'table_caption'},{'bbox':[555,88,684,104],'img_idx':0,'res':[{'confidence':0.9998651146888733,'text':'2023年年度报告','text_region':[[551.0,84.0],[687.0,82.0],[688.0,107.0],[551.0,109.0]]}],'score':0.9478155970573425,'type':'header'},{'bbox':[590,1545,651,1559],'img_idx':0,'res':[{'confidence':0.9947237968444824,'text':'31/211','text_region':[[588.0,1540.0],[654.0,1540.0],[654.0,1565.0],[588.0,1565.0]]}],'score':0.9063873291015625,'type':'footer'}]}| 字段名称 | 数据类型 | 意义 |
| text | str | 文本内容 |
| confidence | float | 文本识别置信度或角度分类置信度 |
| text_region | list | 文本位置坐标 |
| html | str | 表格的html字符 |
| regions | list | 版面分析+表格识别+OCR的结果,每一项为一个list 包含表示区域坐标的 bbox,区域类型的type和区域结果的res三个字段 |
说明:如果需要增加、删除、修改返回字段,可在相应模块的module.py文件中进行修改。
总结
当文本检测、方向分类、文本识别、表格识别和版面识别这几个模型成功串联并发布后,整个系统便能够完整解析PDF、图片等文档中的内容。百度提供的通用OCR模型已经能够满足大部分业务场景中对PDF和图片数据的解析需求,返回结果为JSON格式,其中表格内容则以HTML形式展示。该服务具有基本的商用功能,能够承接文档解析业务。如果再结合APP应用层,该系统可以灵活适应更多业务场景,扩展其应用范围。
此外,通过发布SER(结构化实体识别)模型,系统还能够实现对发票、证件等特殊文档的关键信息抽取,提供更强大的文档解析功能。而该系统架构的最大优势在于支持自定义模型的发布,用户能够基于行业需求对模型进行精调和微调,结合实际业务场景标注数据、训练模型并实现部署,从而大大提升行业专用模型的性能与适配度。
整体来说,这样的架构不仅满足了通用文档解析的需求,还为用户提供了更高的灵活性和可扩展性,使其能够轻松应对复杂多变的业务场景。
| 欢迎光临 链载Ai (https://www.lianzai.com/) | Powered by Discuz! X3.5 |