Commercializing AI Tech | United States Postal Service

Pieces of mail going down conveyor belt to be scanned and sorted.

The United States Postal Service

Artificial intelligence technology enables efficiency
Venu Govindaraju's work in handwriting recognition was at the center of the first handwritten address interpretation system used by the United States Postal Service (USPS). USPS issued a contract to researchers at the University at Buffalo to develop the handwriting recognition technology. One year after implementation it saved the USPS $90 million by automatically processing, and barcoding for precise delivery, more than 25 billion letters. The 2009 Computing Community Consortium dubbed the project as “one of the most successful applications of Machine Learning for developing a real-time engineered system.”

How The USPS Collaborated with UB

Handwriting Recognition Technology     Perfecting Multilingual OCR   |   Enabling Postal Automation

CHALLENGE & OPPORTUNITY

Revolutionizing handwriting understanding & automation

Venu Govindaraju and his team of more than 50 post-doc students and researchers pioneered the world's first self-governing system for handwriting understanding. Operating at 13 postal mail pieces per second, it interprets natural handwriting without strict rules or forms. Govindaraju's method employs advanced algorithms to determine destinations from 170 million possibilities using contextual information and generated lexicons. Overcoming challenges of natural cursive handwriting, they utilized an interactive A*-like algorithm and a stochastic recognizer to cluster and distinguish various writing styles.

Researcher processing documents on laptop with a overlay of cursive writing.
Doctor reviews digital patient record on virtual medical network.

SOLUTION & IMPACT

Expanding AI frontiers

Govindaraju's research impact extends far beyond revolutionizing the postal service industry, branching into Digital Libraries and Multilingual OCR. His methods facilitate early disease detection via healthcare form processing for the New York State (NYS) Department of Health, enhance patient safety through automated reading of medical prescriptions, and enable efficient access to historical documents, including Sanskrit and Arabic texts. Moreover, his techniques enable the retrieval of lecture video segments with significant handwritten and whiteboard content. Through innovations in word spotting, transcript mapping, text retrieval, and writer identification, Govindaraju has introduced powerful methods that advance technology across diverse applications.

RELATED NEWS

Harness AI to build something life-changing. UB can help.

Companies of all sizes work with UB to innovate, solve challenges and bring new products to market. We offer a broad range of services and support for any industry. Our connections make it easy for you to build or grow your business.