Home / Documents / Salary slip

Income

Salary slip OCR API

Read payslips in common layouts and check that the numbers add up: net pay must equal gross earnings minus deductions.

Sample Salary slip (synthetic specimen)
Synthetic specimen. No real person's data.

Real response for this sample

FieldValueConfidence
employerACME TECHNOLOGIES PRIVATE LIMITED99%
month2026-10100%
employee nameRAVI KUMAR100%
employee idACM104299%
designationSenior Engineer100%
panABCPK1234F99%
gross earnings55000.00100%
total deductions2000.00100%
net pay53000.0097%
basic30000.00100%
hra15000.00100%
provident fund1800.00100%
professional tax200.00100%

No review needed. Read in 260 ms.

Fields returned

FieldWhat it is
employer, monthEmployer and pay month
employee_name, employee_id, designationEmployee
gross_earnings, total_deductions, net_payTotals
basic, hra, provident_fund, professional_taxComponents
pan, uan, bank_accountWhere printed

Every field comes with a confidence score from 0 to 1.

Checks

  • Net pay = gross earnings − total deductions
  • Net pay is plausible

If a required field is missing, unsure, or a check fails, the response hasneeds_review: true and a warning explaining why.

Used for

  • Salaried income assessment
  • Personal loans

Request

Send a JPEG, PNG, WebP, TIFF, BMP or PDF up to 10 MB. Leave out document_type to detect it automatically.

curl https://api.lendocr.com/v1/documents:extract \
  -H "Authorization: Bearer lo_test_..." \
  -F file=@document.jpg \
  -F document_type=salary_slip

Authentication, errors, batch uploads and webhooks →

Questions about the salary slip API

What if the salary slip numbers do not add up?

LendOCR checks that net pay equals gross earnings minus deductions. If it does not, the slip is flagged for review with a warning.

Can I send a phone photo of a salary slip?

Yes. Send a JPEG, PNG, WebP, TIFF, BMP or PDF up to 10 MB. Tilted and low-resolution photos are handled; if an image is too unclear, the response is flagged for review and says why.

Do you store the salary slip after reading it?

No. On the extract API the document is processed in memory and dropped. For background jobs it is kept encrypted only until processed, and results are erased after 24 hours.