Image Dark Data Assessment

Dark data are some existing data that the developers or users are unaware of their existence or value, or do not know how to extract value from them. To address the emerging dark data challenge, we proposed a framework for assessing the value of image dark data by using a novel semantic hash ranking (SHR) algorithm. This framework ranked the value of the image dark data, and therefore it helped discover “important images” in the massive image dataset. The evaluation results had demonstrated the performance and validity of our methodology.