Training deep networks for facial expression recognition with crowd-sourced label distribution (2016-08-03T00:00:00.000000Z)

TL;DR

This paper shows how to learn a deep convolutional neural network (DCNN) from noisy labels, using facial expression recognition as an example, and compares four different approaches to utilizing the multiple labels.

Abstract

Crowd sourcing has become a widely adopted scheme to collect ground truth labels. However, it is a well-known problem that these labels can be very noisy. In this paper, we demonstrate how to learn a deep convolutional neural network (DCNN) from noisy labels, using facial expression recognition as an example. More specifically, we have 10 taggers to label each input image, and compare four different approaches to utilizing the multiple labels: majority voting, multi-label learning, probabilistic label drawing, and cross-entropy loss. We show that the traditional majority voting scheme does not perform as well as the last two approaches that fully leverage the label distribution. An enhanced FER+ data set with multiple labels for each face image will also be shared with the research community.

Authors

Zhengyou Zhang

3 papers

Emad Barsoum

1 papers

Cha Zhang

4 papers

TL;DR

Abstract

Authors

References36 items

Increasing the Reliability of Crowdsourcing Evaluations Using Online Quality Assessment

DisturbLabel: Regularizing CNN on the Loss Layer

Deep Residual Learning for Image Recognition

Image based Static Facial Expression Recognition with Multiple Deep Network Learning

Emotion Distribution Recognition from Facial Expressions

Speech-Driven Animation Constrained by Appropriate Discourse Functions

Deeply Learning Deformable Facial Action Parts Model for Dynamic Expression Analysis

CREMA-D: Crowd-Sourced Emotional Multimodal Actors Dataset

Going deeper with convolutions

Very Deep Convolutional Networks for Large-Scale Image Recognition

Facial Expression Recognition via a Boosted Deep Belief Network

Combining modality specific deep neural networks for emotion recognition in video

Challenges in representation learning: A report on three machine learning contests

Deep Learning using Linear Support Vector Machines

ImageNet classification with deep convolutional neural networks

Increasing cheat robustness of crowdsourcing tasks

Classification of Complex Information: Inference of Co-Occurring Affective States from Their Expressions in Speech

Interpreting ambiguous emotional expressions

Cheap and Fast – But is it Good? Evaluating Non-Expert Annotations for Natural Language Tasks

Blur Insensitive Texture Classification Using Local Phase Quantization

Crowdsourcing user studies with Mechanical Turk

The New Handbook of Methods in Nonverbal Behavior Research

Conducting Judgment Studies

Private emotions versus social interaction: a data-driven approach towards analysing emotion in speech

Representing shape with a spatial pyramid kernel

2005 Special Issue: Challenges in real-life emotion annotation and machine learning based detection

Recognizing Action Units for Facial Expression Analysis

Feature-Based Facial Expression Recognition: Sensitivity Analysis and Experiments with A Multilayer Perceptron

Constants across cultures in the face and emotion.

Amazon mechanical turk. https://www.mturk.com

Fer+ emotion label

Facial Action Coding System

Active learning and crowdsourcing for machine translation in low resource scenarios

Crowdsourcing for Affective Annotation of Video: Development of a Viewer-reported Boredom Corpus

Multi-Label Classification of Music into Emotions

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2007 1 Dynamic Texture Recognition Using Local Binary Patterns with an Application to Facial Expressions

Field of Study

Journal Information

Name

Venue Information

Name

Type

URL

Alternate Names