Our Technology
Over the years, Parascript has been established as a trusted provider of state-of-the-art solutions in cognitive computing, applying computer vision and artificial intelligence to a number of markets. Parascript was the first to solve the cursive handwriting puzzle, and over the past 25 years, we’ve grown to be the source for applied artificial intelligence technologies for some of the largest postal services and other Global 2000 organizations.
Our staff—with PhDs in physics and applied mathematics—has decades of experience applying computer vision and AI technologies to complex business problems. From ICR software (that employs a topological approach), recognition that includes curve tracing, feature analysis, and matrix matching all powered by multiple neural networks to powerful Business Intelligence software that learns the proven decision-making practices of a company’s top performers and makes recommendations similar to what these experts would do in the same situations, Parascript has played an active role in the use of learning technologies to tackle these challenges.
The universality of our proprietary technologies, as well as the broad range of expertise and accumulated knowledge allow us to quickly and efficiently apply these technologies to a broad range of fields. For instance, we brought the benefits of exceptional accuracy of region of interest location and low false-positive rates to the medical imaging industry through our AccuDetect CAD for Mammography product.
Key Technologies
Computer Vision
Computer vision involves automating tasks that our own vision accomplishes such as translating digital images into information or detecting certain things. Computer vision is a specialized application of AI in that many of the techniques used to process an image are part of the field of AI and are focused on images.
Typical tasks using computer vision involve object recognition and classification, which includes optical character recognition and handwriting recognition, motion analysis, scene reconstruction and image restoration. These tasks can support controlling roboticized processes, navigation, event detection, organizing information, and automatic inspection.
Examples of computer vision solutions provided by Parascript include:
- Region of interest location on letters, flats and parcels
- Automatic indicia location and detection on envelope images
- Check stock verification and signature verification
- Identification of suspicious areas in medical images such as mammograms
Artificial Intelligence and Machine Learning Algorithms
While artificial intelligence (AI) technologies power our computer vision solutions, applications involving AI go beyond images themselves to any type of digital information. The real excitement of AI is that there is an unprecedented amount of computing power coupled with huge amounts of digital data as input. AI now has both at its disposal in order to learn and quickly compute. The result is that the concept of a true learning machine has become a reality.
Parascript has decades of experience with applying various AI and machine learning algorithms to business problems and combining various algorithms to produce the highest quality results. Each algorithm type has strengths and weaknesses, and therefore, has ideal and less ideal applications. By combining various approaches, outputs can be refined into a synthesized “best answer.”
Among the types of algorithms Parascript uses are convolutional neural networks for deep learning, Conditional Random Fields, Hidden Markov Models, Bayesian-based algorithms, support vector machines and other types of neural networks.
Depending on the particular application, both supervised and unsupervised learning are being utilized.
Examples of applications for machine learning algorithms in Parascript solutions include:
- Document classification
- Predictive analytics in claims data
- Decision support for retail transactions
Every product or solution built by Parascript has at its core multiple AI and computer vision technologies intelligently combined to produce the most reliable, highest accuracy data for your business needs.