Visual Search using Document Page Appearance

Neural Vision Technologies LLC (“NVT”) is a software company with a novel visual search engine which captures the visual appearance of a document page and finds other document pages which are similar in appearance. Highly accurate, it has no reliance on text and is purposefully simple in design. NVT VISTA emulates the way a human sees and remembers what a document looks like.

Most documents have a distinctive visual appearance and are recognizable as being the same record, report, form or template type. For many corporations and law firms, this is a very useful tool for searching large volumes of documents and document pages for record types of interest.

Current methods to search for specific record types rely on key words usually found as part of file name, or by searching the text of document contents. These methods usually fail because file names are imprecise or meaningless and content text searches yield incomplete results or overwhelm users with false positives. Many times, individual documents are combined into a single PDF making their presence opaque.

NVT VISTA yields highly accurate results and leads to more efficient and timely search completions.



NVT VISTA looks at documents the same way that a person does,
using visual signals generated by algorithms and then comparing those signals across a population of documents using deep neural networks. Unlike people, we are very consistent and tireless.


NVT VISTA also supports clustering of the raw document input and automatically groups visually similar documents, allowing the user to understand the gross composition of document types in the collection and use the clusters to build searches.

Who It’s For

NVT VISTA is used by corporations and law firms seeking a better way to find categories of documents.



A defining characteristic of NVT VISTA is extreme simplicity for the user, with a single screen for accessing all functionality. This means training time is minimal and deployments are fast and straightforward.