piXlogic: Analyzing Visual Data to Increase Revenues

Joseph Santucci, President & CEO
Knowledge Management is in part about making sense of data. Historically, these data have comprised structured and unstructured information, where the latter was mostly text data. In today’s world, visual data in the form of images and videos is becoming increasingly important and ubiquitous in many industries. Due to the significant network infrastructure build-up this past decade, corporate communications are also benefiting from increased use of visual media. Companies are discovering that making sense of this data and repurposing it can be a source of competitive advantage and help increase revenue. Unfortunately, mining the content of visual media is a big technical challenge. For the most part, human eyes are needed to make sense of what the image/video contains which is expensive and does not scale well.

Los Altos, CAbased piXlogic has been working for a number of years with government customers in the US and abroad. That experience and the products developed for those environments are well suited to help corporations large and small deal with images and videos in a more effective and scalable way. “Our software solution automatically processes images and videos, creates metadata describing their contents, recognizes and auto-tags faces, text strings, objects, and scenes, and overall helps users find what they need quickly and inexpensively. The software supports a number of critical use cases, and the data generatedis easy to integrate in customer workflows,” says Joseph Santucci, President and CEO, piXlogic.

The company’s software (piXserve) works with over a hundred standard image and video formats and with live video sources. Web sites can also be crawled and indexed. In its most basic implementation, once piXserve has “indexed” a media archive or a live video source, users can query the system by using an image or a text string (ex: here is a picture of a specific car, a picture of a specific person, and the text string “Beach Hotel”; find me images/ video segments where all three are visible in the same shot). Users can also provide an example of something important to them (here is a picture of a “brand x” bar of soap) and ask the software to automatically add the “brand x” keyword every time it processes a picture/video frame where that bar of soap is visible.
These and many other use cases are supported by piXserve. The metadata that piXserve creates can be exported to add value to other customer systems. For example, piXlogic customers in the retail industry have used this capability to process videos of supermarkets shelves and automatically identify and recognize products, where they are located, and how many there are. These data can then be correlated to both cash register data as well as to scheduled planograms: the former to fine-tune marketing/sales campaigns and increase revenue; the latter to reduce compliance costs.

Other customer use cases have revolved around security, industrial inspection, and content discovery. For example, one customer uses piXserve to check defects in wind turbine blades. Video of the blades is collected with small drones. The video is then processed by piXserve. The software is given a set of sample images that define the appearance of the fault conditions of interest. When piXserve goes through the video, it compares against these and the frames where the fault conditions are matched are automatically tagged for review by a specialist inspector. Instead of spending hours reviewing mostly inconsequential video, the inspector spends minutes reviewing only those snippets identified by piXserve as most likely to contain faults. The increase in productivity is huge. In another use case, a customer in the publishing industry put piXserve to use on an archive containing millions of poorly organized and essentially un-tagged photographs. piXserve not only made the material searchable without human intervention, but also generated keyword metadata for use by production workflow systems.

In the security and defense space, piXlogic customers have leveraged piXserve’s ability to detect and recognize objects, faces, and text strings (in a variety of languages), from a range of source-platforms (security cameras, geospatial images, etc.). “piXlogic is very fortunate in that it works with premier partners, such as Carahsoft Technology Corp. These partnerships help piXlogic develop and maintain lasting relationships with its customers,” says Santucci.

Looking forward to the future, piXlogic aspires to empower its end users further, eventually enabling them to interact with the software much the same way as one communicates with a colleague. Working with images and videos is difficult and complex science. piXlogic wants to reduce and shield the user from these complexities thereby enabling broad and widespread enterprise applications.

Company
piXlogic

Headquarters
Los Altos, CA

Management
Joseph Santucci, President & CEO

Description
Provider of software solutions that make media searchable by automatically analyzing and recognizing the contents of images and videos.