首页>
外国专利>
METHOD FOR MULTI-TASK DEEP LEARNING-BASED AESTHETIC QUALITY ASSESSMENT ON NATURAL IMAGE
METHOD FOR MULTI-TASK DEEP LEARNING-BASED AESTHETIC QUALITY ASSESSMENT ON NATURAL IMAGE
展开▼
机译:基于多任务深度学习的自然图像美学质量评估方法
展开▼
页面导航
摘要
著录项
相似文献
摘要
A method for multi-task deep learning-based aesthetic quality assessment on a natural image. The method comprises: step 1, automatically learning about multi-task deep learning-based aesthetic and semantic features of a natural image (S101); and step 2, performing multi-task deep learning-based aesthetic classification and semantic identification on the automatic learning result to implement aesthetic quality assessment on the natural image (S102). The method more efficiently performs aesthetic quality assessment by assisting the expression and learning of aesthetic features using semantic information, and designs a plurality of multi-task deep learning network structures to more efficiently obtain a high-precision classification of an image using aesthetic and semantic information. The method can be applied to many fields related to aesthetic quality assessment of images, including image searching, photography, album management, and the like.
展开▼