Abstract
Reporters obtained the following quote from the background information supplied by the inventors:“Existing methods of image recognition generally performs classification at an object-level. The accuracyin the classification depends on the quantity and quality of training data that is used to train the classifier.In other words, to perform the image recognition accurately, the classifier may need to be trained witha large amount of high quality training data. Further, the computing resources required for running theclassifier may be large, as the classifier may look for features in the entire input image and may attempt tomap these features to a classification type. Further, existing methods of classifying an entire object in animage may result in high error rates, for example, when there are multiple objects with similar appearance.”
NETL
NSTL