FeatherNets: Convolutional Neural Networks as Light as Feather for Face Anti-Spoofing (2019-04-22T00:00:00.000000Z)

TL;DR

An extreme light network architecture(FeatherNet A/B) is proposed with a streaming module which fixes the weakness of Global Average Pooling and uses less parameters and a novel fusion procedure with "ensemble + cascade" structure is presented to satisfy the performance preferred use cases.

Abstract

Face Anti-spoofing gains increased attentions recently in both academic and industrial fields. With the emergence of various CNN based solutions, the multi-modal(RGB, depth and IR) methods based CNN showed better performance than single modal classifiers. However, there is a need for improving the performance and reducing the complexity. Therefore, an extreme light network architecture(FeatherNet A/B) is proposed with a streaming module which fixes the weakness of Global Average Pooling and uses less parameters. Our single FeatherNet trained by depth image only, provides a higher baseline with 0.00168 ACER, 0.35M parameters and 83M FLOPS. Furthermore, a novel fusion procedure with "ensemble + cascade" structure is presented to satisfy the performance preferred use cases. Meanwhile, the MMFD dataset is collected to provide more attacks and diversity to gain better generalization. We use the fusion method in the Face Anti-spoofing Attack Detection Challenge@CVPR2019 and got the result of 0.0013(ACER), 0.999(TPR@FPR=10e-2), 0.998(TPR@FPR=10e-3) and 0.9814(TPR@FPR=10e-4).

Authors

P. Zhang

5 papers

F. Zou

1 papers

Zhiwen Wu

1 papers

TL;DR

Abstract

Authors

References39 items

Bag of Tricks for Image Classification with Convolutional Neural Networks

A Dataset and Benchmark for Large-Scale Multi-Modal Face Anti-Spoofing

MobiFace: A Lightweight Deep Learning Face Recognition on Mobile Devices

Exploiting temporal and depth information for multi-frame face anti-spoofing

ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks

Time Analysis of Pulse-Based Face Anti-Spoofing in Visible and NIR

MobileFaceNets: Efficient CNNs for Accurate Real-time Face Verification on Mobile Devices

Learning Deep Models for Face Anti-Spoofing: Binary or Auxiliary Supervision

Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network

Fd-Mobilenet: Improved Mobilenet with a Fast Downsampling Strategy

ArcFace: Additive Angular Margin Loss for Deep Face Recognition

MobileNetV2: Inverted Residuals and Linear Bottlenecks

Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions

Automatic differentiation in PyTorch

Face anti-spoofing using patch and depth-based CNNs

Squeeze-and-Excitation Networks

Focal Loss for Dense Object Detection

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

Face Antispoofing Using Speeded-Up Robust Features and Fisher Vector Encoding

Understanding the Effective Receptive Field in Deep Convolutional Neural Networks

An original face anti-spoofing approach using partial convolutional neural network

Cross-Database Face Antispoofing with Robust Feature Representation

Xception: Deep Learning with Depthwise Separable Convolutions

Secure Face Unlock: Spoof Detection on Smartphones

Densely Connected Convolutional Networks

Integration of image quality and motion cues for face anti-spoofing: A neural network approach

Deep Residual Learning for Image Recognition

Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification

Do Convnets Learn Correspondence?

Going deeper with convolutions

An Overview of Face Liveness Detection

Network In Network

Context based face anti-spoofing

Can face anti-spoofing countermeasures work in a real world scenario?

Face liveness detection with component dependent descriptor

Face spoofing detection from single images using micro-texture analysis

Dlib-ml: A Machine Learning Toolkit

LBP − T OP based countermeasure against face spooﬁng attacks

Field of Study

Journal Information

Name

Page