In recent years, with the rapid development of computer technology, perception of human facial beauty is an important aspect of human intelligence and has attracted more and more attention of researchers. For the current study methods that exist in the training data set of scoring most depends on manual processes, and the facial beauty assessment is not detailed enough to predict the results, this paper aims to investigate and develop intelligent systems for learning the concept of female facial beauty with data mining learning and producing human-like predictors. Our work is notably different from and goes beyond previous works. We impose less restrictions in terms of pose, lighting, background on the face images used for training and testing, which greatly reduces the manual operation for classification and we do not require costly manual annotation of landmark facial features but simply take raw pixels or texture feature as inputs. We show that a biologically-inspired model with clustering and the improved BP network method can produce results that are much more human-like approach.%随着计算机技术的迅速发展以及人脸识别技术的成熟,人脸美貌度受到越来越多的关注和研究.针对目前的研究方法中存在的对训练数据集的评分过多依赖人工操作,以及对人脸美貌度的预测结果不够详细等问题,本文提出基于HodgeRank的人脸美貌度预测系统,利用数据挖掘方法学习女性人脸的美貌度特征,构造一个模拟预测人脸美貌度的系统.明显区别于之前的研究,该系统训练和测试时采用的人脸数据集放宽了对姿态、光照以及所处环境等条件的限制,评分所需的人工操作大大减少,无需进行大量的人工标定,使用图像的原始像素或纹理特征作为输入,分别采用聚类和改进的BP网络的方法,得到更符合人类特征的美貌度预测结果.
展开▼
