Custom PDF OCR using MERN without typescript
Budget: $250 – $750 USD
Project 1 – Custom OCR
Backend Nodejs and frontend reactjs
Requirement
1. From front end user will upload some set of pdf file in a collection let say 5 pdf file he will upload in a collection A and also by observing the pdf format pattern code will assign one unique identifier also. Like that user can created multiple collection from the fontend. This collection will hold the same format of upload 5 pdf file. For different format user will create another collection B and he will upload another 5 file of same format in collection B. Now he can delete the collection as well or delete the files from the collection. But at max 5 file can be uploaded to that collection for mapping.
2. Once the file is present in the collection user will click on particular collection and all the uploaded file be visible in that collection. There he can select the file and then he can draw the rectangular box in the pdf which is visible and assign some name to that key and also he will drap the box to corresponding value which will be the value of that key and there will be mapping beteen then and this mapping will be stored in the database. Like that there will be multiple mapping.
If the file has table then in case of singleline table the mapping will be as follow
let say table has header, single line body then he will map the header key with the body value
In case of multiline table the mapping will be as follow
let say table has header, multiline body then each key will be mapped to body line so here we get array of object data
At later point user can update the mapping as well which should store in the database
3. Now when the Pdf file containing multiple pages uploaded to the system using api (we need to enable the CORS so that api can be publically accessible using api key – So for each login we are going to assign one unique api key) so user can only access the api with valid api key. So below action will happen
a. user uploaded file will be store on AWS bucket
b. On UI there will be one page where user will see the uploaded file so he will select one pdf file there he can assign the pagenumber for each pages in that pdf then when it click on continue then based on paged number uploaded pdf file be split and all the splited pdf will be upload to AWS again and linking will be store in the database
c. Once the splitting process is completed all the splited file one by one will be passed to the OCR api.
Now for each file OCR will identify the file format from the trained collection automatically then he use use the mapping from that collection and he will extract the data based on mapping and output will be in json format. In one json object all the splitted pdf files data will be there then that data will be store in the database
4. Then we need NLP to query that database
Backend Nodejs and frontend reactjs
Requirement
1. From front end user will upload some set of pdf file in a collection let say 5 pdf file he will upload in a collection A and also by observing the pdf format pattern code will assign one unique identifier also. Like that user can created multiple collection from the fontend. This collection will hold the same format of upload 5 pdf file. For different format user will create another collection B and he will upload another 5 file of same format in collection B. Now he can delete the collection as well or delete the files from the collection. But at max 5 file can be uploaded to that collection for mapping.
2. Once the file is present in the collection user will click on particular collection and all the uploaded file be visible in that collection. There he can select the file and then he can draw the rectangular box in the pdf which is visible and assign some name to that key and also he will drap the box to corresponding value which will be the value of that key and there will be mapping beteen then and this mapping will be stored in the database. Like that there will be multiple mapping.
If the file has table then in case of singleline table the mapping will be as follow
let say table has header, single line body then he will map the header key with the body value
In case of multiline table the mapping will be as follow
let say table has header, multiline body then each key will be mapped to body line so here we get array of object data
At later point user can update the mapping as well which should store in the database
3. Now when the Pdf file containing multiple pages uploaded to the system using api (we need to enable the CORS so that api can be publically accessible using api key – So for each login we are going to assign one unique api key) so user can only access the api with valid api key. So below action will happen
a. user uploaded file will be store on AWS bucket
b. On UI there will be one page where user will see the uploaded file so he will select one pdf file there he can assign the pagenumber for each pages in that pdf then when it click on continue then based on paged number uploaded pdf file be split and all the splited pdf will be upload to AWS again and linking will be store in the database
c. Once the splitting process is completed all the splited file one by one will be passed to the OCR api.
Now for each file OCR will identify the file format from the trained collection automatically then he use use the mapping from that collection and he will extract the data based on mapping and output will be in json format. In one json object all the splitted pdf files data will be there then that data will be store in the database
4. Then we need NLP to query that database