Computer vision and deep OCR in the enterprise: 3 use cases

Description

The use of deep learning in the field of computer vision is a relatively new concept, with the first major breakthrough coming in 2012 during the ImageNet competition. That was the first time a deep learning network outperformed traditional computer vision techniques, and it won by a landslide. The ImageNet challenge was to classify images of objects (i.e., balloons, cats, etc.) across ~20 thousand categories. Even today, years after the first use of deep learning in computer vision, the kind of problems that are commonly studied is the classification of objects that are easily differentiable visually (i.e., person versus tree). Dave Ferrell explores three examples of nontraditional techniques pushing the boundaries of computer vision in industries today, including identifying “unseen” objects.