information community file
Technical capability
Voice technology
Character recognition
Face and Human Body
Image technology
Language and knowledge
video technique

Test paper analysis and identification

Interface description

It can analyze the document layout, output the positions of graphs, tables, titles, and texts, and output the OCR recognition results of the content in different sections. It supports Chinese and English languages, multiple scenarios of mixed handwriting and print, and formula recognition and handwritten vertical recognition.

Online debugging

You can visit Sample Code Center Debug the interface in , you can perform signature verification, view the request content and return results of online calls, and automatically generate sample code.

Request Description

Request Example

HTTP method: POST

Request URL: https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis

URL parameter:

parameter value
access_token Access_token obtained through API Key and Secret Key, refer to“ Access Token acquisition

The headers are as follows:

parameter value
Content-Type application/x-www-form-urlencoded

Place the request parameters in the body. The details of the parameters are as follows:

Request Parameters

parameter Required type Optional value range explain
image And url/pdf_file string - For image data, the size of urlencode after base64 encoding and urlencode shall not exceed 10M, the shortest side shall be at least 15px, the longest side shall be at most 8192px, and jpg/jpeg/png/bmp format is supported
priority : image>url>pdf_file, when the image field exists, the url and pdf_file fields become invalid
url And image/pdf_file string - The image is a complete url with a length of no more than 1024 bytes. The size of the image corresponding to the url after encoding in base64 does not exceed 10M. The shortest side is at least 15px, and the longest side is at most 8192px. It supports the jpg/jpeg/png/bmp format
priority : image>url>pdf_file, when the image field exists, the url field is invalid
Please close the URL anti-theft chain
pdf_file And image/url string - PDF files are urlencoded after base64 encoding. It is required that the size of base64 encoding and urlencoded files should not exceed 10M, the shortest side should be at least 15px, and the longest side should be at most 8192px
priority : image>url>pdf_file, when the image and url fields exist, the pdf_file field is invalid
pdf_file_num no string - The corresponding page number of the PDF file that needs to be identified. When the pdf_file parameter is valid, identify the corresponding page content of the incoming page number. If not, identify the first page by default
language_type no string CHN_ENG/ ENG Recognition language type, default is CHN_ENG
Optional values include:
=CHN_ENG: Chinese and English
=ENG: English
result_type no string big/small Whether the recognition result is returned as a single line result or a single word result, the default is big.
=Big: Return the line identification result
=Small: returns a single word result in addition to the line recognition result
detect_direction no string true/false Whether to detect the orientation of the image, the default is not to detect, that is, false. Orientation means that the input image is in the normal direction and rotates 90/180/270 degrees counterclockwise. Among them,
0: Forward
1: Rotate 90 degrees counterclockwise
2: Rotate 180 degrees counterclockwise
3: Rotate 270 degrees counterclockwise
line_probability no string true/false Whether to return the confidence level of recognition results of each line. Default is false
disp_line_poly no string true/false Whether to return the four corner coordinates of each line. Default is false
words_type no string handwring_only/ handprint_mix Text type.
Default: printed text recognition
=Handwriting_only: handwritten character recognition
=Handprint_mix: handwritten printing mixed arrangement recognition
layout_analysis no string true/false Analyze document layout: including layout (figure, table, title, paragraph, and table of contents); Analysis output of attribute (column, header, footer, page number, footnote)
recg_formula no string true/false Whether to detect and recognize the formula. The default value is false. The formula is returned in Latex format text.
=True: Detect and recognize formulas
=False: Do not detect recognition formula
recg_long_division no string true/false Whether to detect and recognize the handwriting vertical type. The default value is false.
=True: Detect and recognize handwriting vertical
=False: Do not detect handwriting vertical

Request Code Example

Prompt 1 : Before using the sample code, remember to replace the sample token, image address or Base64 information.

Prompt 2 : Some languages depend on classes or libraries. Please check the download address in the code comment.

 #Test paper analysis and identification
 curl -i -k ' https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis?access_token= [Call the token obtained from the authentication interface] ' --data 'language_type=CHN_ENG&result_type=big&image=[Picture Base64 encoding, UrlEncode required]' -H 'Content-Type:application/x-www-form-urlencoded'
 # encoding:utf-8

 import requests import base64 ''' Test paper analysis and identification ''' request_url =  " https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis "
 #Open picture file in binary mode f =  open ( '[Local file]' ,  'rb' ) img = base64 . b64encode ( f . read ( ) ) params =  { "image" : img , "language_type" : "CHN_ENG" , "result_type" : "big" } access_token =  '[Token obtained by calling the authentication interface]' request_url = request_url +  "?access_token="  + access_token headers =  { 'content-type' :  'application/x-www-form-urlencoded' } response = requests . post ( request_url , data = params , headers = headers )
 if response :
     print  ( response . json ( ) )
 package  com . baidu . ai . aip ;

 import  com . baidu . ai . aip . utils . Base64Util ;
 import  com . baidu . ai . aip . utils . FileUtil ;
 import  com . baidu . ai . aip . utils . HttpUtil ;

 import  java . net . URLEncoder ;

 /** *Document layout analysis and recognition */
 public  class  DocAnalysis  {

     /** *Tool class required in important tip code *FileUtil, Base64Util, HttpUtil, GsonUtils *  https://ai.baidu.com/file/658A35ABAB2D404FBF903F64D47C1F72 *  https://ai.baidu.com/file/C8D81F3301E24D2892968F09AE1AD6E2 *  https://ai.baidu.com/file/544D677F5D4E4F17B4122FBD60DB82B3 *  https://ai.baidu.com/file/470B3ACCA3FE43788B5A963BF0B625F3 *Download */
     public  static  String  docAnalysis ( )  {
         //Request url
         String url =  " https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis " ;
         try  {
             //Local file path
             String filePath =  [Local file path] ;
             byte [ ] imgData =  FileUtil . readFileByBytes ( filePath ) ;
             String imgStr =  Base64Util . encode ( imgData ) ;
             String imgParam =  URLEncoder . encode ( imgStr ,  "UTF-8" ) ;

             String param =  "language_type="  +  "CHN_ENG"  +  "&result_type="  +  "big"  +  "&image="  + imgParam ;

             //Note that the purpose here is to simplify the encoding and obtain access_token for each request. The online environment access_token has an expiration time, and the client can cache it and retrieve it after expiration.
             String accessToken =  "[Token obtained by calling the authentication interface]" ;

             String result =  HttpUtil . post ( url , accessToken , param ) ;
             System . out . println ( result ) ;
             return result ;
         }  catch  ( Exception e )  { e . printStackTrace ( ) ;
         }
         return  null ;
     }

     public  static  void  main ( String [ ] args )  {
         DocAnalysis . docAnalysis ( ) ;
     }
 }
 # include  <iostream>
 # include  <curl/curl.h>

 //Download link of libcurl library: https://curl.haxx.se/download.html
 //Download link of jsoncpp library: https://github.com/open-source-parsers/jsoncpp/
 const  static std :: string request_url =  " https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis " ;
 static std :: string docAnalysis_result ;
 /** *The curl sends the callback function called by the http request. The returned body in json format is parsed in the callback function, and the parsing result is stored in the global static variable *See the libcurl document for @ param parameter definitions *@ return See the libcurl document for the definition of the return value */
 static size_t callback ( void  * ptr , size_t size , size_t nmemb ,  void  * stream )  {
     //The obtained body is stored in ptr and converted to string format first docAnalysis_result = std :: string ( ( char  * ) ptr , size * nmemb ) ;
     return size * nmemb ;
 }
 /** *Document layout analysis and recognition *@ return If the call is successful, 0 will be returned. If an error occurs, other error codes will be returned */
 int  docAnalysis ( std :: string & json_result ,  const std :: string & access_token )  { std :: string url = request_url +  "?access_token="  + access_token ; CURL * curl =  NULL ; CURLcode result_code ;
     int is_success ; curl =  curl_easy_init ( ) ;
     if  ( curl )  {
         curl_easy_setopt ( curl , CURLOPT_URL , url . data ( ) ) ;
         curl_easy_setopt ( curl , CURLOPT_POST ,  one ) ; curl_httppost * post =  NULL ; curl_httppost * last =  NULL ;
         curl_formadd ( & post ,  & last , CURLFORM_COPYNAME ,  "language_type" , CURLFORM_COPYCONTENTS ,  "CHN_ENG" , CURLFORM_END ) ;
         curl_formadd ( & post ,  & last , CURLFORM_COPYNAME ,  "result_type" , CURLFORM_COPYCONTENTS ,  "big" , CURLFORM_END ) ;
         curl_formadd ( & post ,  & last , CURLFORM_COPYNAME ,  "image" , CURLFORM_COPYCONTENTS ,  "【base64_img】" , CURLFORM_END ) ;

         curl_easy_setopt ( curl , CURLOPT_HTTPPOST , post ) ;
         curl_easy_setopt ( curl , CURLOPT_WRITEFUNCTION , callback ) ; result_code =  curl_easy_perform ( curl ) ;
         if  ( result_code != CURLE_OK )  {
             fprintf ( stderr ,  "curl_easy_perform() failed: %s\n" ,
                     curl_easy_strerror ( result_code ) ) ; is_success =  one ;
             return is_success ;
         } json_result = docAnalysis_result ;
         curl_easy_cleanup ( curl ) ; is_success =  zero ;
     }  else  {
         fprintf ( stderr ,  "curl_easy_init() failed." ) ; is_success =  one ;
     }
     return is_success ;
 }
 <? php
 /** *Initiate http post requests (REST APIs) and obtain the results of REST requests * @param string $url * @param string $param * @return - http response body if succeeds, else false. */
 function  request_post ( $url  =  '' ,  $param  =  '' )
 {
     if  ( empty ( $url )  ||  empty ( $param ) )  {
         return  false ;
     }

     $postUrl  =  $url ;
     $curlPost  =  $param ;
     //Initialize curl
     $curl  =  curl_init ( ) ;
     curl_setopt ( $curl ,  CURLOPT_URL ,  $postUrl ) ;
     curl_setopt ( $curl ,  CURLOPT_HEADER ,  zero ) ;
     //The result is required to be a string and output to the screen
     curl_setopt ( $curl ,  CURLOPT_RETURNTRANSFER ,  one ) ;
     curl_setopt ( $curl ,  CURLOPT_SSL_VERIFYPEER ,  false ) ;
     //Post submission method
     curl_setopt ( $curl ,  CURLOPT_POST ,  one ) ;
     curl_setopt ( $curl ,  CURLOPT_POSTFIELDS ,  $curlPost ) ;
     //Run curl
     $data  =  curl_exec ( $curl ) ;
     curl_close ( $curl ) ;

     return  $data ;
 }

 $token  =  '[Token obtained by calling the authentication interface]' ;
 $url  =  ' https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis?access_token= '  .  $token ;
 $img  =  file_get_contents ( '[Local file path]' ) ;
 $img  =  base64_encode ( $img ) ;
 $bodys  =  array (
     'language_type'  = >  "CHN_ENG" ,
     'result_type'  = >  "big" ,
     'image'  = >  $img
 ) ;
 $res  =  request_post ( $url ,  $bodys ) ;

 var_dump ( $res ) ;
 using System ;
 using System . IO ;
 using System . Net ;
 using System . Text ;
 using System . Web ;

 namespace com . baidu . ai {
     public  class  DocAnalysis
     {
         //Document layout analysis and recognition
         public  static  string  docAnalysis ( )
         {
             string token =  "[Token obtained by calling the authentication interface]" ;
             string host =  " https://aip.baidubce.com/rest/2.0/ocr/v1/doc_analysis?access_token= "  + token ;
             Encoding encoding = Encoding . Default ;
             HttpWebRequest request =  ( HttpWebRequest ) WebRequest . Create ( host ) ; request . Method =  "post" ; request . KeepAlive =  true ;
             //Base64 encoding of pictures
             string base64 =  getFileBase64 ( [Local picture file] ) ;
             String str =  "language_type="  +  "CHN_ENG"  +  "&result_type="  +  "big"  +  "&image="  + HttpUtility . UrlEncode ( base64 ) ;
             byte [ ] buffer = encoding . GetBytes ( str ) ; request . ContentLength = buffer . Length ; request . GetRequestStream ( ) . Write ( buffer ,  zero , buffer . Length ) ;
             HttpWebResponse response =  ( HttpWebResponse ) request . GetResponse ( ) ;
             StreamReader reader =  new  StreamReader ( response . GetResponseStream ( ) , Encoding . Default ) ;
             string result = reader . ReadToEnd ( ) ; Console . WriteLine ( "Document layout analysis and recognition:" ) ; Console . WriteLine ( result ) ;
             return result ;
         }

         public  static  String  getFileBase64 ( String fileName )  {
             FileStream filestream =  new  FileStream ( fileName , FileMode . Open ) ;
             byte [ ] arr =  new  byte [ filestream . Length ] ; filestream . Read ( arr ,  zero ,  ( int ) filestream . Length ) ;
             string baser64 = Convert . ToBase64String ( arr ) ; filestream . Close ( ) ;
             return baser64 ;
         }
     }
 }

Return description

Return parameters

field Required type explain
log_id yes uint64 Unique log ID for problem location
img_direction no int32 Detect_direction=true. Detected image orientation, 0: positive direction; 1: Rotate 90 degrees counterclockwise; 2: Rotate 180 degrees counterclockwise; 3: Rotate 270 degrees counterclockwise
results_num yes uint32 Number of recognition results, representing the number of elements of results
results yes array[] Identification result array
+ words_type yes string Text attributes (handwriting, printing), handwriting, printing
+ words yes array[] The recognition result array of the whole line.
++ line_probability no array[] It is returned when line_probability=true. The confidence value of each line in the recognition result, including average: average value of line confidence, min: minimum value of line confidence
+++ average no float Row confidence
+++ min no float The lowest confidence level of a word in the whole line
++ word yes string Identification result of the whole line
++ poly_location no array[] Whether to return the coordinates of the four corners of each line, when disp_line_poly=true
++ words_location yes array[] The rectangular box coordinates of the whole line. Location information (coordinate 0 is the upper left corner)
+++ left yes uint32 The horizontal coordinate of the top left vertex of the rectangle representing the positioning position
+++ top yes uint32 The vertical coordinate of the top left vertex of the rectangle representing the positioning position
+++ width yes uint32 The width of the rectangle representing the positioning position
+++ height yes uint32 Height of rectangle representing position
+ chars no array[] Result_type=small. Single character result array
++ char no string Result_type=small. Content of each word
++ chars_location no object The rectangular box coordinates of each word. Location information (coordinate 0 is the upper left corner)
+++ left no uint32 The horizontal coordinate of the top left vertex of the rectangle representing the positioning position
+++ top no uint32 The vertical coordinate of the top left vertex of the rectangle representing the positioning position
+++ width no uint32 The width of the rectangle representing the positioning position
+++ height no uint32 Height of rectangle representing position
formula_result no array[] Identify the formula array in the result, including the formula location and formula content,
When recg_formal=true
+ form_location no array[] The rectangular box coordinate array of the formula in the recognition result (coordinate 0 point is the upper left corner)
+ form_words no string Identify the content of the formula in the result
words_result no array[] The recognition result array after the fusion of ordinary text and formula,
When recg_formal=true
+ location no array[] The rectangular box coordinate array of the whole line in the recognition result (coordinate 0 point is the upper left corner)
+ words no string Identify the contents of the whole line in the result
+ chars no array[] Single character result array. The formula as a whole is a single word,
Result_type=small
++ char no string Content of each word
++ chars_location no object Rectangular box coordinate array of each word (coordinate 0 point is the upper left corner)
layouts_num no uint32 Number of layout analysis results, representing the number of layout elements
layouts no array[] The document layout module array in each "column: section" contains five modules, including table, figure, paragraph text, title, and directory; Coordinate position of each module; The row serial number id corresponding to the paragraph text and the text content in the table.
+ layout no string Label results of layout analysis. Table: table, figure, text, title, contents
+ layout_location no array[] The position of the document layout information label, four vertices: top left, top right, bottom right, bottom left
++ x no uint32 Horizontal coordinate (coordinate 0 is the upper left corner)
++ y no uint32 Horizontal coordinate (coordinate 0 is the upper left corner)
+ layout_idx no array[] The position of the text in the document layout information in the results: if the row serial number ID corresponding to the layout text label is n, the text in this label will be displayed in the n+1 item in the results)
sec_rows no uint32 Show the "column: section" content in all layouts as a grid of M x N, sec_rows=M
sec_cols no uint32 Show the "column" content in all layouts as a grid of M x N, sec_cols=N
sections no array[] The five page attributes contained in a picture include: column, header, footer, page number, and footer. The array contains the attribute label, attribute location, and ID number of the text content contained in the attribute.
Among them, the section contains five module contents, including table, figure, paragraph text, title and directory (output in the return parameter layouts)
+ attribute no string Attribute label results of layout analysis, column: section, header: header, footer: footer, page number: number, footnote: footnote
+ attri_location no array[] The location of layout analysis attributes, four vertices: top left, top right, bottom right, bottom left
++ x no uint32 Horizontal coordinate (coordinate 0 is the upper left corner)
++ y no uint32 Horizontal coordinate (coordinate 0 is the upper left corner)
+ sec_idx no string Sections returns the serial number identification of the contents contained in the five layout attributes in the parameter
++ idx no string Sections returns the serial number of the text line ID contained under each of the five layout attributes in the parameter
++ para_idx no string It will be returned only when attribute=section. Indicates the sequence number id returned by the five modules including table, figure, paragraph text, title and directory in the "Column: section" of the return parameter (that is, the return sequence number of each module in the returned results of layouts)
++ row_idx no string It will be returned only when attribute=section. Indicates that all columns are represented as M xN grids, and the ID of the grid row
++ col_idx no string It will be returned only when attribute=section. Indicates that all columns are represented as M xN grids, and the column ID of the grid
+ long_division no array[] Handwritten vertical recognition result, returned when recg_long_division=true
+ location no object Handwritten vertical rectangular box coordinate array (coordinate 0 point is the upper left corner)
+ words no object Output handwritten vertical inner text results by line
++ word no string Content of each line of text
++ words_location no object Rectangular box coordinate array of each line (coordinate 0 point is the upper left corner)
+ long_division_num no uint32 The number of handwritten vertical recognition results, representing the number of long_division elements
pdf_file_size no string The total number of pages of the incoming PDF file. This field is returned when the pdf_file parameter is valid

Return to Example

 {
	 "results_num" :  six ,
	 "log_id" :  "4488766695474114139" ,
	 "img_direction" :  zero ,
	 "layouts_num" :  zero ,
	 "results" :  [
		 {
			 "words_type" :  "print" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  one hundred and twenty-four ,
					 "left" :  one hundred and thirty-six ,
					 "width" :  four hundred and eighteen ,
					 "height" :  sixty-five
				 } ,
				 "word" :  "Five dictations (4 points)"
			 } ,
		 } ,
		 {
			 "words_type" :  "print" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  two hundred and forty-six ,
					 "left" :  one hundred and thirty-six ,
					 "width" :  thirty-seven ,
					 "height" :  forty-five
				 } ,
				 "word" :  "1"
			 } ,
		 } ,
		 {
			 "words_type" :  "handwriting" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  one hundred and ninety-five ,
					 "left" :  two hundred and thirty-seven ,
					 "width" :  four hundred and sixty-nine ,
					 "height" :  one hundred and four
				 } ,
				 "word" :  "Picking chrysanthemums under the east fence"
			 } ,
		 } ,
		 {
			 "words_type" :  "print" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  two hundred and forty-one ,
					 "left" :  eight hundred and eighty-nine ,
					 "width" :  two hundred and eighty-seven ,
					 "height" :  fifty-two
				 } ,
				 "word" :  "See Nanshan leisurely?"
			 } ,
		 } ,
		 {
			 "words_type" :  "print" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  four hundred and fifteen ,
					 "left" :  one hundred and thirty-four ,
					 "width" :  four hundred and seventy-two ,
					 "height" :  fifty-two
				 } ,
				 "word" :  "2. Businesswomen don't know the hatred of national subjugation"
			 } ,
		 } ,
		 {
			 "words_type" :  "handwriting" ,
			 "words" :  {
				 "words_location" :  {
					 "top" :  three hundred and seventy-seven ,
					 "left" :  six hundred and seven ,
					 "width" :  five hundred and fifty-six ,
					 "height" :  ninety-three
				 } ,
				 "word" :  "Across the river you still sing backyard flowers."
			 } ,
		 } ,
	 ] ,
   "formula_result" :  [
         {
             "form_location" :  {
                 "top" :  zero ,
                 "left" :  ninety-seven ,
                 "width" :  one hundred and fifty-one ,
                 "height" :  seventy-seven
             } ,
             "form_words" :  " x = \\frac { 1 } { n - 1 } - 1 1 \\frac { \\frac { 5 } { 2 } } { 5 }"
         } ,
         {
             "form_location" :  {
                 "top" :  one hundred and nineteen ,
                 "left" :  one hundred and eighteen ,
                 "width" :  one hundred and fifteen ,
                 "height" :  eighty
             } ,
             "form_words" :  " = \\sqrt { \\frac { x } { 2 } ( x - 1 ) ^ { 2 } }"
         } ,
         {
             "form_location" :  {
                 "top" :  one hundred and ninety-six ,
                 "left" :  seventy-eight ,
                 "width" :  seventeen ,
                 "height" :  twenty-four
             } ,
             "form_words" :  " x ^ { 2 }"
         } ,
         {
             "form_location" :  {
                 "top" :  two hundred and forty-four ,
                 "left" :  seventy-nine ,
                 "width" :  one hundred and three ,
                 "height" :  seventy
             } ,
             "form_words" :  " s = \\frac { \\sum _ { i = 0 } { m } \\cdot i v } { - 1 }"
         }
     ] ,
     "words_result" :  [
         {
             "location" :  {
                 "top" :  one hundred and sixty-four ,
                 "left" :  two hundred and fifty-five ,
                 "width" :  one hundred and eleven ,
                 "height" :  sixteen
             } ,
             "words" :  "Where m represents the examinee"
         } ,
         {
             "location" :  {
                 "top" :  one hundred and ninety-eight ,
                 "left" :  twenty-four ,
                 "width" :  three hundred and forty-one ,
                 "height" :  eighteen
             } ,
             "words" :  "The number of people x ^ {2} represents the equal score of the first question in the exam."
         } ,
     ] ,
 }
Previous
Character recognition of medical bills
Next
Formula recognition