AntWorks CMR+
by AntWorks Pte Ltd
CMR+ uses AI and Machine Learning to extract data from text & image documents with high accuracy.
AntWorks is a global AI company with market presence across US, UK, Europe, Mumbai, Singapore and Tokyo with R&D centers in India at Chennai and Bangalore.
AntWorks CMR+ extracts data from a variety of documents including handwritten application forms, contract documents, financial agreements and identity documents.
- image-preprocessing that allows for data cleansing, skew correction, watermark removal, noise removal and orientation correction
- Image sorting- a typical docket would comprise of multiple document types necessitating classification of individual documents for appropriate handling
- extraction- here the user has granular control to extract a variety of information form the document including header content (name, invoice number etc), tabular content (item names, pricing, shipping dates etc) along with object details like stamps, seals, signatures etc
- Business rules- this would performing data transformation, validation and conversion of data into required structure across json and xml formats.
The results are presented to the quality-assurance team for validation. The feedback is leveraged by the algorithm for improving the extraction over time. These include computer vision algorithms, NLPs for semantic language understanding along with supervised learning algorithms for auto-correction, feedback learning and region detection.
The platform has been built to be enterprise-grade based on the following
- Low-code-no-code platforms to enable business technologists with a Software Development Toolkit(SDK) to stretch the boundaries of extraction capability to achieve industry leading accuracy
- Sophistication to address process-specific needs including workflows, custom output formats etc
- Security based on deep encryption standards and granular access controls to meet regulated industry needs (banking, healthcare etc)
- Integration capability into enterprise systems with flexible output structures & native RPA
Use cases
- Application forms- including handwriting extraction, check boxes, signature, tables, stamps,
- Semi-structured documents like invoices, POs, bill of ladings, cheque etc
- extraction of language from contracts, mortgage documents
- Identity documents- driver license, passports
The product has been deployed across US, UK, Europe, India, Singapore and Japan for use cases including insurance claims, contract comparison, trade finance, invoice processing, Libor repapering.