PatentImages is a visual discovery engine for patent drawings. For example, if you're a maker of digital pianos and want to know what's been patented for weighted piano keys, you simply input the phrase "weighted piano keys" and ranked ordered tiles of relevant patent drawings are returned within seconds.
Unless you are a patent attorney or professional patent searcher, most people do not revel the thought of reading a patent let alone multiple patents. However patent searching and freedom to operate is an essential step before commercialising and launching new products. The patent system has been in place for at least 200 years with the goal of recognising and rewarding inventors for disclosure of their inventions. Whilst not a perfect system, it has been the best system available to promote innovation which society has been the beneficiary of for the last 200 years.
Since the digitisation of patent documents in the early 21st century, the way to search for patents has been to input a text-based search query and then to manually review the search results which are a list of patents and their full text (abstract, description and claims).
PatentImages is different - its image based alongside plain language summaries of the patent that ordinary people can use to understand what the patent is really about and determine whether they need to explore further or discard as irrelevant. Our AI classifies and segments patent drawings and generates a context vector that is mapped to token-based semantic vectors of the patent text.
PatentImages is part of the Berserq ecosystem. A useful signal is provided to us in deidentified form pertaining to technology areas that are trending.