ibml & ScaleHub recognize and extract handwritten text with 99.x% accuracy
In a recent interview published by two German magazines, our Chief Revenue Officer, Torsten Malchow, and Steffen Unmuth, Sales Director at ibml, discussed our combined effort to provide scalable, affordable process automation that doesn’t require internal resources for data capture.
Forms filled out by hand have long been an Achilles heel of process automation efforts. Document scanning service providers and mailrooms still have to rely on people to interpret handwriting and manually key the information into company systems. But as Steffen Unmuth explained, “thanks to cloud technology and the support of crowd workers, we’ve been able to develop a service for handwriting recognition.”
Perhaps most impressive, the service promises (and delivers) an unprecedented 99.x% rate of accuracy. “Since we still have to rely on people to interpret handwritten documents,” said Steffen, “it’s the way those tasks are organized that makes the difference. With our service, businesses tap into the power of crowdsourcing and cloud computing to extract and digitize handwritten content contained in documents.”
Here’s how it works: from scanned documents, the solution takes the fields filled out by hand and breaks them into snippets, tags each snippet with a unique code and consolidates similar snippets before dispersing them as tasks to crowd workers around the world. The crowd workers log in via secure web browser to view handwritten snippets and manually key the information into a text field.
“Scalability is our greatest advantage,” said Steffen. “Not only does this service allow production managers to complete mass amounts of work in very little time, but also scale up and down whenever needed; for example, in response to seasonal fluctuations. Our crowdsourcing solution accesses more than 2.3 million crowd workers worldwide.”
To get up and running, ibml customers just need to add a Crowdconnector module to their CaptureSuite, and data extraction jobs are automatically forwarded to crowd workers via the cloud. Within minutes, the solution delivers recognized and extracted data back to CaptureSuite
Torsten Malchow explained: “The combination of AI and human intelligence is what differentiates our approach from traditional “intelligent automation” providers. Self-learning, intelligent capture components for automated data extraction, when paired with crowd workers who manually capture the snippets of data that cannot be automatically captured, is the key to decreasing complexity and lowering costs.”