Extracting data from payment screenshot like gpay, phone pay, yono, paytm etc

Help Activities
1

Hello everyone, Keshav here. Currently, I am working on a project where I have to extract data from the payment screenshots like GPay, PhonePay, Paytm, etc., and after extracting, I have to make an Excel to store all the details like sender name, date, time, amount, etc. After doing this, I have to match the payment transactions with the bank statement. Is it possible through UiPath Studio? I have tried a lot by using the Extract document Data prompt based on data extratction but the problem is that it is not 100% accurate. I worked on many transactions and matched also, but not for every image, for 50 screenshots, data is extracted, but sometimes it doesn’t. Sometimes it changes the date and time, or even sometimes it doesn’t reads amount. Is there any possible way to do this.?

Forum Engagement Daily Reports
2

Hello @Keshav_Choudhari

I am not a developer, but i have an idea how to implement this.

But success rate i am assume 60 - 70% remaining you can use human loop or some manual reconciliation

Here is my approach will be,

Classify the Screenshot (GPay / PhonePe / Paytm / Bank App / UPI)

Different apps → different formats.

Use:

UiPath Document Understanding – *Document Classifier
or
Build a Prompt-Based Classifier (GenAI)
Ex: Identify whether this is a GPay, PhonePe, Paytm, or Bank payment screenshot. Return only the app name.

Use App-Specific OCR and Extraction

Each app always follows the same pattern for date, time, amount, reference ID, etc.

OmniPage OCR (highest accuracy)
Microsoft Read OCR
Tesseract OCR (as backup)

Then run Regex to extract reliable fields.
Even if OCR misreads:
*13 becomes 18
2024 → 2029
₹690 becomes ₹600
You can fix it using a Validation Prompt Ex:
Extract: sender name, amount, UPI ID, date, time, transaction ID.
Correct common OCR errors (e.g., digit confusion).
Return a valid JSON.

Write All Records to Excel then Match Data with Bank Statement

Hope it give helps

3

@Keshav_Choudhari

You can use, Read PDF with OCR activity and using Regex you can get all the details that you required.

One more way to get the solution of this, use document understanding to extract all these details. If you dont have an idea on this go with 1st solution, if you want try document understading with different types extractor that cant extract with 1st solution you can use this DU.

Go through academy,

If you find this helpful please mark it as solution.

Thanks,
YK.