Deep Learning Face Attributes in the Wild

Abstract

Predicting face attributes in the wild is challenging due to complex face variations. We propose a novel deep learning framework for attribute prediction in the wild. It cascades two CNNs, LNet and ANet, which are fine-tuned jointly with attribute tags, but pre-trained differently. LNet is pre-trained by massive general object categories for face localization, while ANet is pre-trained by massive face identities for attribute prediction. This framework not only outperforms the state-of-the-art with a large margin, but also reveals valuable facts on learning face representation. (1) It shows how the performances of face localization (LNet) and attribute prediction (ANet) can be improved by different pre-training strategies. (2) It reveals that although the filters of LNet are fine-tuned only with image-level attribute tags, their response maps over entire images have strong indication of face locations. This fact enables training LNet for face localization with only image-level annotations, but without face bounding boxes or landmarks, which are required by all attribute recognition works. (3) It also demonstrates that the high-level hidden neurons of ANet automatically discover semantic concepts after pre-training with massive face identities, and such concepts are significantly enriched after fine-tuning with attribute tags. Each attribute can be well explained with a sparse linear combination of these concepts.

To appear in International Conference on Computer Vision (ICCV) 2015

References in corpus (4)

• Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition
• Fully Convolutional Networks for Semantic Segmentation
• Deep Attribute Networks
• Self-taught Object Localization with Deep Networks

Cited by in corpus (54)

• Mixed Precision Training
• Improving Person Re-identification by Attribute and Identity Learning
• A Richly Annotated Dataset for Pedestrian Attribute Recognition
• Self-supervised learning of a facial attribute embedding from video
• Unsupervised Image-to-Image Translation with Generative Adversarial Networks
• Co-training for Demographic Classification Using Deep Learning from Label Proportions
• Adversarial Examples in Modern Machine Learning: A Review
• Recurrent Generative Adversarial Networks for Proximal Learning and Automated Compressive Image Recovery
• ST-GAN: Spatial Transformer Generative Adversarial Networks for Image Compositing
• Pseudo-task Augmentation: From Deep Multitask Learning to Intratask Sharing---and Back
• Scribbler: Controlling Deep Image Synthesis with Sketch and Color
• Deep Learning For Face Recognition: A Critical Analysis
• Nonlinear 3D Face Morphable Model
• Rethinking Feature Distribution for Loss Functions in Image Classification
• Foreground-aware Image Inpainting
• Gang of GANs: Generative Adversarial Networks with Maximum Margin Ranking
• Attribute-Guided Face Generation Using Conditional CycleGAN
• Beyond Shared Hierarchies: Deep Multitask Learning through Soft Layer Ordering
• Precomputed Real-Time Texture Synthesis with Markovian Generative Adversarial Networks
• The Riemannian Geometry of Deep Generative Models
• FacePoseNet: Making a Case for Landmark-Free Face Alignment
• Triple consistency loss for pairing distributions in GAN-based face synthesis
• Facelet-Bank for Fast Portrait Manipulation
• Disentangling Factors of Variation by Mixing Them
• Deep Semantic Face Deblurring
• Class Rectification Hard Mining for Imbalanced Deep Learning
• FAN: Feature Adaptation Network for Surveillance Face Recognition and Normalization
• Fine-grained Synthesis of Unrestricted Adversarial Examples
• A Domain Based Approach to Social Relation Recognition
• GANHopper: Multi-Hop GAN for Unsupervised Image-to-Image Translation
• Face Translation between Images and Videos using Identity-aware CycleGAN
• Recent Advances of Image Steganography with Generative Adversarial Networks
• Learning Disentangling and Fusing Networks for Face Completion Under Structured Occlusions
• Deep learning-based Real-time Volumetric Imaging for Lung Stereotactic Body Radiation Therapy: A Proof of Concept Study
• Commutative Lie Group VAE for Disentanglement Learning
• MR-GAN: Manifold Regularized Generative Adversarial Networks
• PixelNN: Example-based Image Synthesis
• SSSE: Efficiently Erasing Samples from Trained Machine Learning Models
• Unsupervised 3D Shape Learning from Image Collections in the Wild
• Geometry-Aware Face Completion and Editing
• Measuring Fairness in Generative Models
• Imbalanced Deep Learning by Minority Class Incremental Rectification
• Faceness-Net: Face Detection through Deep Facial Part Responses
• Murine AI excels at cats and cheese: Structural differences between human and mouse neurons and their implementation in generative AIs
• Deep Network Interpolation for Continuous Imagery Effect Transition
• Unpaired Multi-Domain Image Generation via Regularized Conditional GANs
• Improving Bi-directional Generation between Different Modalities with Variational Autoencoders
• FaceShapeGene: A Disentangled Shape Representation for Flexible Face Image Editing
• Introspective Generative Modeling: Decide Discriminatively
• Deep Imbalanced Attribute Classification using Visual Attention Aggregation
• Wasserstein Introspective Neural Networks
• A Generative Map for Image-based Camera Localization
• Deception Detection by 2D-to-3D Face Reconstruction from Videos
• Vehicle Image Generation Going Well with The Surroundings