2021 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW) 2021
DOI: 10.1109/iccvw54120.2021.00394
|View full text |Cite
|
Sign up to set email alerts
|

Prior Aided Streaming Network for Multi-task Affective Analysis

Abstract: Automatic affective recognition has been an important research topic in human computer interaction (HCI) area. With recent development of deep learning techniques and large scale in-the-wild annotated datasets, the facial emotion analysis is now aimed at challenges in the real world settings. In this paper, we introduce our submission to the 2nd Affective Behavior Analysis in-the-wild (ABAW2) Competition. In dealing with different emotion representations, including Categorical Emotions (CE), Action Units (AU),… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
4
0

Year Published

2022
2022
2024
2024

Publication Types

Select...
5
2
2

Relationship

1
8

Authors

Journals

citations
Cited by 22 publications
(4 citation statements)
references
References 48 publications
0
4
0
Order By: Relevance
“…This approach highlights the effectiveness of bifurcated networks in handling complex facial recognition tasks in varied conditions. Continuing this trend, Zhang et al [ 76 ] combine datasets Aff-Wild2, BP4D, and DFEW, utilizing a CNN for feature extraction and obtaining a 79.3% accuracy rate. Shabbir et al [ 90 ] propose the Fine-Grained Bilinear CNN (FgbCNN) to enhance the generalization of facial features in expressions such as anger and fear, achieving high accuracies on multiple datasets.…”
Section: Resultsmentioning
confidence: 99%
“…This approach highlights the effectiveness of bifurcated networks in handling complex facial recognition tasks in varied conditions. Continuing this trend, Zhang et al [ 76 ] combine datasets Aff-Wild2, BP4D, and DFEW, utilizing a CNN for feature extraction and obtaining a 79.3% accuracy rate. Shabbir et al [ 90 ] propose the Fine-Grained Bilinear CNN (FgbCNN) to enhance the generalization of facial features in expressions such as anger and fear, achieving high accuracies on multiple datasets.…”
Section: Resultsmentioning
confidence: 99%
“…We verify the effectiveness of our proposed prior embedding module and streaming model respectively. For the prior embedding module, we conduct ablation study and prove that our method not only concentrates on the facial areas mostly and conveys human emotions, but also ignores the noise areas like backgrounds 12 . For the streaming module, we compare our method with/without the AU→EXPR, AU→VA, and EXPR→VA modules.…”
Section: Automatic Affective Recognitionmentioning
confidence: 97%
“…For example, in affective analysis, human emotion comes with varying representation forms, including category expression, action units and valence/arousal. They are inherently related, and based on this, several multi-tasking frameworks have been exploited [166], [167]. Besides, in [168], the authors simultaneously predicted the systolic and diastolic blood pressure from PPG, achieving better performance than directly estimating each individual parameter.…”
Section: Learn With Supplementary Tasksmentioning
confidence: 99%