Automating the Extraction of Useable Knowledge From Videos and Photos

Photo credit: Kiyoshi Takahase Segundo

Do you have a need to index and search pictures and video? If you are working for any of the intelligence agencies, a police department, or any other security force that involves video and photo, you may. You may even have a need for this outside of the intelligence/security realm. How do you do it now? Does a human need to look at each video and picture and index what objects are visible in the video? That could take ages if so, and may be nearly impossible to get through the large amounts of data.

Right now there are several companies that can index and search video in different ways, but most do not offer a general Google-like search of video and pictures. Additionally, most of these products are not scalable for enterprise class purposes. Some of the companies out there include Numenta, Pixsy, IDEE TinEye, and KEVA, which is only available to government currently.

California based piXlogic has brought an interesting product to the table that is changing the way enterprises search video and photos. piXserve is an enterprise class 64bit Java application for Windows. It indexes all objects in video and photos and attempts to classify the objects. It does this by breaking a picture down into smaller pieces and creating a specific mathematical description of that image to leave metadata that can be searched. piXserve extracts small parts of a picture or video (in any format) and creates a 3D object rotation in space, filling in the blanks about what the oject would look like from other angles. piXserve works for many use cases, but pays specific attention to facial recognition and text recognition. It also works very well with geospatial imagery.

In one demo, piXserve took a picture of from a freeze frame of video. It extracted information from the face of a passerby and searched other videos in the database for the same person. It actually picked up many other videos of the same person, including videos looking at the same person from different perspectives. I could see this product having great application to law enforcement and intelligence agencies specifically, especially in cities with a large amount of cameras around the city. It could be highly beneficial for investigative and forensic purposes. What other areas do you see this product being useful?

To watch the Carahsoft webinar about this product, click here.


Original post

Leave a Comment

Leave a comment

Leave a Reply