ShownX / FacePaperCollection

A collection of face related papers

Face Related Papers and Code Collection

Any face research/engineer related merge request is wellcome! 02/08/2019.

Index

1. Toolkits
2. Face Detection
   • Survey
   • Datasets
   • Research
3. Face Alignment
   • Survey
   • Datasets
   • Research
4. Face Recosntruction
   • Survey
   • Datasets
   • Research
5. Face Recognition
   • Survey
   • Tutorial
   • Datasets
   • Template Generator
     • Pretrained models
     • Image-based Template Generator
     • Image-set-based Template Generator
   • Face Recognition Pipeline
6. Face Generation
   • Survey
   • Datasets
   • Research
7. Face Attributes Analysis
   • Survey
   • Datasets
   • Research

Toolkits

• FaRE: Open Source Face Recognition Performance Evaluation Package [Paper] [Code is coming soon!]
• Gluon Toolkit for Face Recognition [MXNET]
• Deep Learning:
  • MXNet and Gluon: A flexible and efficient library for deep learning.
  • Torch and PyTorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration.
  • TensorFlow: An open-source software library for Machine Intelligence.
  • Caffe and Caffe2: A lightweight, modular, and scalable deep learning framework.
• Machine Learning:
  • Dlib: A machine learning toolkit.
• Computer Vision:
  • OpenCV: Open Source Computer Vision Library.
• Probabilistic Programming
  • Pyro: Deep universal probabilistic programming with Python and PyTorch

Face Detection

Survey

Datasets

• Wildest Faces: Face Detection and Recognition in Violent Settings
• WIDER FACE: A Face Detection Benchmark [Project]
• FDDB: Face Detection and Data Set Benchmark [Project]
• AFLW: Annotated Facial Landmarks in the Wild: A Large-scale, Real-world Database for Facial Landmark Localization [Project]

Research

• PyramidBox: A Context-assisted Single Shot Face Detector [ Paper] [TensorFlow] [PyTorch] [MXNet]
• Face Attention Network: An Effective Face Detector for the Occluded Faces [Paper] [PyTorch]
• FaceNess-Net: Face Detection through Deep Facial Part Responses: [Paper]
• S³FD: Single Shot Scale-invariant Face Detector [Paper] [Caffe] [PyTorch]
• Finding Tiny Faces: [Project] [Paper] [MatConvNet + MATLAB] [TensorFlow] [MXNET]
• SSH: Single Stage Headless Face Detector: [Paper] [Caffe] [TensorFlow] [MXNET]
• Focal Loss for Dense Object Detection: [Paper] [Caffe] [TensorFlow] [MXNET]
• Face R-CNN: [Paper] [Caffe]
• FaceBoxes: A CPU Real-time Face Detector with High Accuracy [Paper] [Caffe]
• Multiview Face Detection: [Paper] [Caffe]

Face Alignment

Survey

Datasets

• LS3D-W: How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks) [Project]
• AFLW: Annotated Facial Landmarks in the Wild: A Large-scale, Real-world Database for Facial Landmark Localization. [Project]
• 300-W [Project]
• 300-VW [Project]

Research

• FAN: How far are we from solving the 2D & 3D Face Alignment problem? [Paper] [PyTorch]
• JFA: Joint Head Pose Estimation and Face Alignment Framework Using Global and Local CNN Features [Paper]
• MDM: Mnemonic Descent Method [Paper] [TensorFlow]
• RDL: Recurrent 3D-2D Dual Learning for Large-pose Facial Landmark Detection [Paper]
• PIFA: Pose-invariant 3D face alignment [Paper] [Code]

Face Reconstruction

Survey

Datasets

Research

• UH-E2FAR: End-to-end 3D face reconstruction with deep neural networks: [Paper]
• Multi-View 3D Face Reconstruction with Deep Recurrent Neural Networks: [Paper]
• 3D Face Morphable Models "In-the-Wild" [Paper]
• 3DMM-CNN [Paper] [Code]
• VRN [Paper] [Code] [Online Demo]
• 3DFaceNet [Paper]
• MoFA: Unsupervised learning for 3D model and pose parameters [Paper]
• 3DMM-STN: Using 3DMM to transfer 2D image to 2D image texture [Paper]
• Dense Semantic and Topological Correspondence of 3D Faces without Landmarks
• Generating 3D Faces using Convolutional Mesh Autoencoders [Paper] [Code]

Face Recognition

Survey

Tutorial

• Deep Learning for Face Recognition

Datasets

Training sets:

• MS-Celeb-1M: Microsoft dataset contains around 1M subjects [Project] [Paper]
• CASIA WebFace: 10,575 subjects and 494,414 images [Project] [Paper]
• CelebA: 202,599 images and 10,177 subjects, 5 landmark locations, 40 binary attributes [Project]
• VGG-Face2: A large-scale face dataset contains 3.31 million imaes of 9131 identities. [Project]

Face Verification

• LFW: Labeled Face in the Wild: 13,000 images and 5749 subjects [Download]
• CFP: Celebrities in Frontal-Profile in the Wild [Project] [Paper]
• MegaFace: 1 Million Faces for Recognition at Scale, 690,572 subjects [Download]
• Surveillance Face Recognition Challenge [Project] [Paper]

Face Closed-set Identification

• UHDB31: UHDB31: A Dataset for Better Understanding Face Recognition across Pose and Illumination Variation [Paper]

Face Open-set Identification

• IJB-C: IARPA Janus Benchmark-C: Face dataset and protocol [Paper]
• IJB-B: IARPA Janus Benchmark-B Face Dataset [Paper]
• IJB-A: Pushing the frontiers of unconstrained face detection and recognition: IARPA Janus Benchmark A [Paper]
• Unconstrained Face Detection and Open-Set Face Recognition Challenge [Project] [Paper]
• MegaFace: 1 Million Faces for Recognition at Scale, 690,572 subjects [Download]

Template Generators

Pretrained Models

• ResNet-101, DenseNet-121 provided by FaRE
• ResNet-50, SE-ResNet-50 provided by VGG-Face2 [Download]
• VGG-16 provided by VGG-Face
• InsightFace [Download]

Image-based Template Genearator

• Pairwise Relation Network, ECCV18: [Paper]
• GridFace: Face Rectification via Learning Local Homography Transformation, ECCV18: [Paper]
• Consensus-Driven Propagation in Massive Unlabeled Data for Face Recognition, ECCV18: [Paper]
• Face Recognition with Contrastive Convolution, ECCV18: [Paper]
• FaceNet: A Unified Embedding for Face Recognition and Clustering, CVPR15 [Paper] [TensorFlow]
• DeepID series, CVPR14: [DeepID] [DeepID2] [DeepID3]
• DeepFace: Closing the Gap to Human-Level Performance in Face Verification, CVPR14: [Paper]

Image-set-based Template Generator

• Dependency-aware Attention Control for Unconstrained Face Recognition with Image Sets, ECCV, 2018
• Comparator Network, ECCV, 2018 [Pytorch]

Training Loss

• InsightFace (ArcFace): Additive Angular Margin Loss for Deep Face Recognition, ArXiv, 2018 [MXNet]
• CosFace: Large Margin Cosine Loss for Deep Face Recognition, CVPR, 2018 [TensorFlow] [MXNet]
• Ring loss: Convex Feature Normalization for Face Recognition [Paper] [PyTorch]
• Git Loss for Deep Face Recognition [Paper]
• A-Softmax Loss (SphereFace) [Paper] [Caffe] (Caffe)
• Triplet Loss [Paper] [Torch] [TensorFlow]
• Center Loss [Paper] [Caffe + MATLAB] [MXNet]
• Range Loss [Paper] [Caffe]
• L-Softmax [Paper] [Caffe] [MXNet]
• Marginal Loss [Paper]

Face Recognition Pipeline

• UR2D-E:Evaluation of a 3D-aided Pose Invariant 2D Face Recognition System
• SeetaFaceEngine: An open source C++ face recognition engine. [C++]
• OpenFace: Face recognition with Google's FaceNet deep neural network using Torch] [Torch +Python]

Face Genearation

Survey

Datasets

Research

1. TP-GAN: [Paper]
2. FF-GAN: [Paper]
3. DR-GAN: [Paper] [Website]
4. BEGAN: Boundary Equilibrium Generative Adversarial Networks [Paper]

Face Attributes Analysis

Survey

Datasets

Research