Client storyAI-enabled Document Operations · Torrent Power
Using AI to classify business documents
Machine learning and OCR applied to document identification and classification.
Results reported by SoftAge
Figures relate to this published project. Read the original SoftAge case study ↗
Managing records for an electricity business
Torrent Power works across electricity generation, transmission and distribution. This project focused on identifying and organising documents more efficiently.
Project overview
SoftAge combined document processing with AI and machine learning to support Torrent Power’s record-management requirements.
The challenge
The workflow needed to distinguish between three form types, identify relevant areas and classify scanned records.
How SoftAge approached it
- 01
Build the workflow
The solution used OCR, OpenCV and Tesseract within the document-processing framework.
- 02
Train the models
The published approach describes iterative training using approximately 5,000–10,000 samples.
- 03
Test classification
Testing checked the system’s output against the project’s requirements.
Training around the documents
The team worked with three form types, identified relevant fields and used OCR with machine learning for classification. Iterative training and testing helped assess the output against the project requirements.
Need help with a similar business process?
Tell us about your processes, documents or data requirements.