Methods and systems for testing an optical character recognition model
AssigneeThe Boeing Company
InventorsAli Azad, David D. Pokrajac, Amir Sadrpour, Barnabas Poczos, Hai Thanh Pham
Life of this patent
Estimating the term…
Abstract
Systems and methods for generating optical character recognition (OCR) models configured to identify characters from a variety of different documents. The OCR models are based on a base model. One or more outside models can be tested to determine their effectiveness in supplementing the base model. When an outside model is effective it is incorporated into the base model. Generations of base models can be created that provide for additional functionality that is not present in the preceding model. A family of the generations of base models is maintained.
Claims20
A method of testing an optical character recognition (OCR) model, the method comprising:
- determining a criteria for improvement of an OCR base model;
- establishing a communication link through a communications network;
- downloading through the communications network an outside model that provides an OCR function;
- obtaining test data comprising control images having predetermined alphanumeric data;
- testing the outside model with the control images;
- based on the testing, determining that the outside model increases the criteria of the base model;
- and
- selecting the outside model for incorporation with the OCR base model.
A computing device comprising:
- interface circuitry configured to send and receive data through a communications network;
- processing circuitry operatively coupled to the interface circuitry and configured to:
- obtain an outside model through the communications network that provides an OCR function;
- obtain test data that comprises control images with known OCR attributes;
- test the outside model using the control images and determine a score;
- when the score is below a predetermined level, determine that the outside model should not be incorporated into a OCR base model;
- and
- when the score is above a predetermined level, generate a new generation base model that comprises the outside model and a base model.
A non-transitory computer readable medium storing a computer program product for controlling a programmable computing device, the computer program product comprising software instructions that, when executed on processing circuitry of the programmable computing device, cause the processing circuitry to:
- establish a communication link through a communications network;
- download through the communications network an outside model that provides an OCR function;
- obtain test data comprising control images having predetermined alphanumeric data;
- test the outside model with the control images;
- based on the testing, determine that the outside model improves at least one criteria of an OCR base model;
- and
- select the outside model for incorporation with the OCR base model.
Dependencies are inferred from the claim text.