BeautyNet: Joint Multiscale CNN and Transfer Learning Method for Unconstrained Facial Beauty Prediction

Abstract

Because of the lack of discriminative face representations and scarcity of labeled training data, facial beauty prediction (FBP), which aims at assessing facial attractiveness automatically, has become a challenging pattern recognition problem. Inspired by recent promising work on fine-grained image classification using the multiscale architecture to extend the diversity of deep features, BeautyNet for unconstrained facial beauty prediction is proposed in this paper. Firstly, a multiscale network is adopted to improve the discriminative of face features. Secondly, to alleviate the computational burden of the multiscale architecture, MFM (max-feature-map) is utilized as an activation function which can not only lighten the network and speed network convergence but also benefit the performance. Finally, transfer learning strategy is introduced here to mitigate the overfitting phenomenon which is caused by the scarcity of labeled facial beauty samples and improves the proposed BeautyNet’s performance. Extensive experiments performed on LSFBD demonstrate that the proposed scheme outperforms the state-of-the-art methods, which can achieve 67.48% classification accuracy.

Skin Lesion Classification Based on Deep Convolutional Neural Networks Architectures

“…7 illustrates key differences between conventional machine learning and transfer learning. Transfer learning can address small datasets, leading to augmenting model learning performance [61,62]. Fine-tuning of trained models allows adaptation to problems to be achieved [63].…”

Facial Beauty Prediction and Analysis based on Deep Convolutional Neural Network: A Review

“…It achieves an effective and simple feature fusion technique through ECCV HotOrNot and SCUT-FBP datasets. Zhai et al [46] showed promising results on fine-grained image classification using a multi-scale architecture to increase diversification among the deep features using BeautyNet to predict unconstrained facial beauty. A multi-scale network was used to enhance the discriminative features of the face.…”

“…A small portion of beautiful faces exists in databases, with most of the faces generated from confined ethnic groups. Previous findings have shown that deeper representation has higher efficient performance than inadequate ones [46,57]. The solution to this inadequacy is to employ state-of-the-art techniques such as data augmentation or transfer learning so as to improve the most and least attractive faces [21], [22], [47].…”

Photofeeler-D3: A Neural Network with Voter Modeling for Dating Photo Impression Prediction

“…Datasets There are a variety of benchmark datasets for scoring images: The AVA dataset [27], the Hot-Or-Not dataset [19], the SCUT-FBP dataset [16], the LSFCB dataset [20], the London Faces Dataset [13], and the CelebA dataset [28]. The AVA dataset [27] doesn't have attractiveness ratings for the subject, instead they have an attractiveness rating for the entire image i.e.…”