Semantic Image Segmentation for Efficiently Adding Recognition Objects

초록

With the development of artificial intelligence technology, various methods have been developed forrecognizing objects in images using machine learning. Image segmentation is the most effective among thesemethods for recognizing objects within an image. Conventionally, image datasets of various classes are trainedsimultaneously. In situations where several classes require segmentation, all datasets have to be trainedthoroughly. Such repeated training results in low training efficiency because most of the classes have alreadybeen trained. In addition, the number of classes that appear in the datasets affects training. Some classes appearin datasets in remarkably smaller numbers than others, and hence, the training errors will not be properlyreflected when all the classes are trained simultaneously. Therefore, a new method that separates some classesfrom the dataset is proposed to improve efficiency during training. In addition, the accuracies of theconventional and proposed methods are compared.

키워드