Discriminant Distribution-Agnostic Loss for Facial Expression Recognition in the Wild

Farzaneh, Amir Hossein; Qi, Xiaojun

doi:10.1109/cvprw50498.2020.00211

Cited by 53 publications

(16 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…PhaNet (Liu et al 2019) 54.82 ESR-9 (Siqueira et al 2020) 59.30 RAN (Wang et al 2020b) 59.50 SCN (Wang et al 2020a) 60.23 PSR (Vo et al 2020) 60.68 EfficientFace (Zhao et al 2021) 59.89 EfficientNet-B0 (Savchenko 2021) 61.32 MViT (Li et (Chen et al 2019) 61.25 LDL-ALSG (Chen et al 2020) 59.35 VGG-FACE (Kollias et al 2020) 60.00 OADN (Ding et al 2020) 61.89 DDA-Loss (Farzaneh et al 2020) 62.34 EfficientFace (Zhao et al 2021) 63.70 MViT (Li et al 2021) 64.57…”

Section: Methodsmentioning

confidence: 99%

“…In recent years, several studies demonstrate that discriminative loss functions could be well adapted to the FER task. (Farzaneh et al 2020) combine the advantages of center loss and softmax loss and propose a DDA loss. Concretely, a center loss aggregates the features of the same class into a cluster, and a softmax loss separates the adjacent classes.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Distract Your Attention: Multi-head Cross Attention Network for Facial Expression Recognition

Wen¹,

Lin²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

We present a novel facial expression recognition network, called Distract your Attention Network (DAN). Our method is based on two key observations. Firstly, multiple classes share inherently similar underlying facial appearance, and their differences could be subtle. Secondly, facial expressions exhibit themselves through multiple facial regions simultaneously, and the recognition requires a holistic approach by encoding high-order interactions among local features. To address these issues, we propose our DAN with three key components: Feature Clustering Network (FCN), Multi-head cross Attention Network (MAN), and Attention Fusion Network (AFN). The FCN extracts robust features by adopting a large-margin learning objective to maximize class separability. In addition, the MAN instantiates a number of attention heads to simultaneously attend to multiple facial areas and build attention maps on these regions. Further, the AFN distracts these attentions to multiple locations before fusing the attention maps to a comprehensive one. Extensive experiments on three public datasets (including Affect-Net, RAF-DB, and SFEW 2.0) verified that the proposed method consistently achieves state-of-the-art facial expression recognition performance. Code will be made available at: https://github.com/yaoing/DAN.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Distract Your Attention: Multi-head Cross Attention Network for Facial Expression Recognition

Wen¹,

Lin²,

Wang³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Accuracy (%) DLP-CNN [57] 74.20 FSN [60] 72.46 DDA loss [61] 79.71 ALT [62] 76.50 Separate loss [63] 77.25 PSR [59] 80.78 Our Method 78.57…”

Section: Methodsmentioning

confidence: 99%

Vision-Based Road Rage Detection Framework in Automotive Safety Applications

Leone

Caroppo

Manni

et al. 2021

Sensors

View full text Add to dashboard Cite

Drivers’ road rage is among the main causes of road accidents. Each year, it contributes to more deaths and injuries globally. In this context, it is important to implement systems that can supervise drivers by monitoring their level of concentration during the entire driving process. In this paper, a module for Advanced Driver Assistance System is used to minimise the accidents caused by road rage, alerting the driver when a predetermined level of rage is reached, thus increasing the transportation safety. To create a system that is independent of both the orientation of the driver’s face and the lighting conditions of the cabin, the proposed algorithmic pipeline integrates face detection and facial expression classification algorithms capable of handling such non-ideal situations. Moreover, road rage of the driver is estimated through a decision-making strategy based on the temporal consistency of facial expressions classified as “anger” and “disgust”. Several experiments were executed to assess the performance on both a real context and three standard benchmark datasets, two of which containing non-frontal-view facial expression and one which includes facial expression recorded from participants during driving. Results obtained show that the proposed module is competent for road rage estimation through facial expression recognition on the condition of multi-pose and changing in lighting conditions, with the recognition rates that achieve state-of-art results on the selected datasets.

show abstract

“…Li et al [38] proposed Deep Locality-Preserving CNN being trained using Locality-Preserving loss to enforce the intra-class compactness by locally clustering deep features using the k-nearest neighbor algorithm. Discriminant distribution-agnostic loss [39] enforces the inter-class dissimilarity which can be useful while dealing with extremely imbalanced datasets. Hayale et al [40] proposed an algorithm for automated FER to preserve the local structure of images in the embedding similarity space.…”

Section: Related Workmentioning

confidence: 99%

Ad-Corre: Adaptive Correlation-Based Loss for Facial Expression Recognition in the Wild

Fard

Mahoor

2022

IEEE Access

View full text Add to dashboard Cite

Automated Facial Expression Recognition (FER) in the wild using deep neural networks is still challenging due to intra-class variations and inter-class similarities in facial images. Deep Metric Learning (DML) is among the widely used methods to deal with these issues by improving the discriminative power of the learned embedded features. This paper proposes an Adaptive Correlation (Ad-Corre) Loss to guide the network towards generating embedded feature vectors with high correlation for within-class samples and less correlation for between-class samples. Ad-Corre consists of 3 components called Feature Discriminator, Mean Discriminator, and Embedding Discriminator. We design the Feature Discriminator component to guide the network to create the embedded feature vectors to be highly correlated if they belong to a similar class, and less correlated if they belong to different classes. In addition, the Mean Discriminator component leads the network to make the mean embedded feature vectors of different classes to be less similar to each other. We use Xception network as the backbone of our model, and contrary to previous work, we propose an embedding feature space that contains k feature vectors. Then, the Embedding Discriminator component penalizes the network to generate the embedded feature vectors, which are dissimilar. We trained our model using the combination of our proposed loss functions called Ad-Corre Loss jointly with the cross-entropy loss. We achieved a very promising recognition accuracy on AffectNet, RAF-DB, and FER-2013. Our extensive experiments and ablation study indicate the power of our method to cope well with challenging FER tasks in the wild.

show abstract

Discriminant Distribution-Agnostic Loss for Facial Expression Recognition in the Wild

Cited by 53 publications

References 25 publications

Distract Your Attention: Multi-head Cross Attention Network for Facial Expression Recognition

Distract Your Attention: Multi-head Cross Attention Network for Facial Expression Recognition

Vision-Based Road Rage Detection Framework in Automotive Safety Applications

Ad-Corre: Adaptive Correlation-Based Loss for Facial Expression Recognition in the Wild

Contact Info

Product

Resources

About