Unlock the intelligence within your data.
AI powered document processing and automated data extraction.
How it works
Zanran's AI recognises layout as humans do. Quickly recognising text, graphs, images, tables and content.
Zanran recognises the elements on a page and automatically extracts them.
The content on the page is automatically available in multiple formats including Excel, XML, JSON, HTML and images
Zanran Workbench provides a powerful visualisation environment to interact with the document and all its content (tables, images, etc). This environment includes an impressive list of productivity tools: Automatic content tagging, On-board calculator, Table editing and exporting, Archiving, User annotator, etc.
Zanran automatically validates the tables in documents by checking every calculation (total, sub-total, etc) using powerful AI and heuristic algorithms. Errors and anomalies are colour coded and display in the Workbench where built-in validation tools help the user to manually verify calculations.
AI DEEP LEARNERS
AI deep learner technology is at the core of many of Zanran’s product features.
Within Zanran’s engine, one finds different deep learner models solving different tasks, but they can be categorised in two groups: deep learners dealing with computer vision tasks (layout analysis, table detection, image classification, etc) and those dealing with language tasks (textual content labelling, etc).
Zanran uses the backbone network called Feature Pyramid Net, built on top of ResNet which computes feature maps of an image in order to solve vision-related tasks.
LANGUAGE BASED DEEP LEARNERS
Zanran uses various deep learner language models including BERT transformers which are trained to learn language representations and act as the basis for natural language processing (NLP) within Zanran's solutions.