Background and Challenges
As a leading domestic commercial bank, the bank has a wide range of business categories and paper document styles. The traditional OCR recognition system cannot meet the business's expected accuracy rate, so it can only rely on manual entry to digitalize paper documents. Under the background of rising costs, the bank urgently needs to introduce internationally leading computer vision technology to liberate low-end labor and improve operating efficiency.
For various types of bank documents, we aim to design out a character recognition system with a recognition rate not lower than that of humans in order to achieve the goal of replacing low-end labor and improving business efficiency.
Difficulties and Key Success Factors
In the huge amount of handwritten documents, everyone's writing method is different, so how to improve the accuracy rate and how to combine with business processes are the key success factors to achieve the landing of the text recognition model.
Through the 4paradigm Sage modeling platform, modeling experts and business experts have designed a brand-new business system and a highly accurate, automatically updateable identification model, which greatly exceeds the expectations of customers and the baseline of landing requirements.
In the project bidding stage, the 4Paradigm recognition model is more than twice powerful that of the opponent. The cutting-edge text recognition technology provided by the 4Paradigm not only meets the recognition needs of various types of text, standard fonts, handwriting, etc., but also has an accuracy rate that can compare or even surpass human recognition rate.
Based on such exciting business outcomes, the bank will continue to work with the 4Paradigm to explore the deeper application of artificial intelligence technology in the field of computer vision, including broader scenarios, more diverse recognition needs to provide better service experience to customers.