Contextual Targeting for Video: There's a New Kid in Town

Seems like the crowding field of contextual video ad vendors and networks greets a new contender about every other month. This month it’s Digitalsmiths, which has come out with a content matching product called VideoSense. The company claims it targets ads based on three factors: frame-by-frame visual analysis of a clip, indexing of the clip’s audio or soundtrack, and analysis of the text and meta-data around the video.

Other firms working toward the same goal include ScanScout, Blinkx and Nexidia. Most use a network model but some are licensing to publishers. The leap of faith with these very cool sounding products and ad networks lies in the image recognition technology. In the case of VideoSense, DigitalSmiths claims to combine “facial recognition, object recognition, scene classification and other contextual algorithms.” However, my conversations with vendors and agency folks indicate these functions are under ripe, and most targeting comes down to audio and meta-data analysis. (release)