In this photo from a leading womens retail website, the model’s skin exposure, body angle and facial positioning is detected and output by GumGum’s LURID image classifying system along with a comprehensive score of image safety and an estimate of what she’s wearing. As seen here, the image is classified as (generally) safe.
How would your computer know if the content of an image is safe? Can computer intelligence correctly detect levels of safe and unsafe image content?
These are some of the questions our team of image scientists have been working to answer, and have resulted in some of the core technologies that differentiate GumGum from other image-based advertising providers.
Led by our director of image technology, Daniel S. Rosen, GumGum’s patented technology breaks down an image to its core elements using several heuristic methods, including image science, human physiology and plain old common sense analytics.
Through this approach, Rosen and his team created a system capable of detecting skin tones, facial and body measurements/positions, which collectively constitute GumGum’s LURID image classifier- a revolutionary system that generates an output score along with analytical data determining whether an image is generally safe, generally unsafe, probably safe or probably unsafe.
Providing this level of analysis and detection leaves the subjective decision making of whether to display an ad on a particular image in the hands of advertisers and account managers, instead of allowing machines to decide.