Automated Depression Diagnosis Based on Facial Dynamic Analysis and Sparse Coding

Wen, Lingyun; Li, Xin; Guo, Guodong; Zhu, Yu

doi:10.1109/tifs.2015.2414392

Cited by 106 publications

(59 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RMSE MAE Baseline [22] 13.61 10.88 LPQ + SVR [45] 10.82 8.97 PHOG [30] 10.45 N/A MHH + EOH + LBP [25] 11.19 9.14 LPQ-TOP + MFA [33] 10.27 8.22 LPQ + Geo [46] 9.72 7.86 Two DCNN [10] 9.82 7.58 C3D Tight-Face [44] 9.64 7.50 C3D Loose-Face [44] 10.04 8.15 RNN-C3D Combined Faces [44] 9.28 7.37 C3D Global + 3D-GAP (Ours) 9.24 7.10 C3D Local + 3D-GAP (Ours) 8.37 6.51 C3D Global and Local + 3D-GAP (Ours) 8.26 6.40 [36] 10.50 8.44 LGBP-TOP + LPQ [37] 10.27 8.20 Two DCNN [10] 9.55 7.47 VGG + FDHH [38] 8.04 6.68 C3D Tight-Face [44] 9.66 7.48 C3D Loose-Face [44] 9.81 7.73 RNN-C3D Combined Faces [44] 9.20 7.22 C3D Global + 3D-GAP (Ours) 8.97 7.09 C3D Local + 3D-GAP (Ours) 8.55 6.81 C3D Global and Local + 3D-GAP (Ours) 8.31 6.59…”

Section: Methodsmentioning

confidence: 99%

“…Cummins et al [30] investigated the use of two different descriptors which are called Space-Time Interest Points (STIP) [31] and Pyramid of Histogram of Gradients (PHOG) [32], with PHOG demonstrating better accuracy. Lingyun Wen et al [33] proposed extracting dynamic feature descriptors based on LPQ from Three Orthogonal Planes (LPQ-TOP). Sparse coding is then applied to organize the feature descriptors, and SVR allows predicting the levels of depression level.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Combining Global and Local Convolutional 3D Networks for Detecting Depression from Facial Expressions

Melo

Granger

Hadid

2019

2019 14th IEEE International Conference on Automatic Face &Amp; Gesture Recognition (FG 2019)

View full text Add to dashboard Cite

Deep learning architectures have been successfully applied in video-based health monitoring, to recognize distinctive variations in the facial appearance of subjects. To detect patterns of variation linked to depressive behavior, deep neural networks (NNs) typically exploit spatial and temporal information separately by, e.g., cascading a 2D convolutional NN (CNN) with a recurrent NN (RNN), although the intrinsic spatio-temporal relationships can deteriorate. With the recent advent of 3D CNNs like the convolutional 3D (C3D) network, these spatio-temporal relationships can be modeled to improve performance. However, the accuracy of C3D networks remain an issue when applied to depression detection. In this paper, the fusion of diverse C3D predictions are proposed to improve accuracy, where spatio-temporal features are extracted from global (full-face) and local (eyes) regions of subject. This allows to increasingly focus on a local facial region that is highly relevant for analyzing depression. Additionally, the proposed network integrates 3D Global Average Pooling in order to efficiently summarize spatio-temporal features without using fully-connected layers, and thereby reduce the number of model parameters and potential over-fitting. Experimental results on the Audio Visual Emotion Challenge (AVEC 2013 and AVEC 2014) depression datasets indicates that combining the responses of global and local C3D networks achieves a higher level of accuracy than state-of-the-art systems.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Combining Global and Local Convolutional 3D Networks for Detecting Depression from Facial Expressions

Melo

Granger

Hadid

2019

2019 14th IEEE International Conference on Automatic Face &Amp; Gesture Recognition (FG 2019)

View full text Add to dashboard Cite

show abstract

“…In Table 2, the performance of our proposed method is compared with state-of-the-art methods on AVEC2013 dataset. The methods in [17,18,19,20,21] are based on handcrafted features. For instance, Local Phase Quantization (LPQ) is employed as baseline method in AVEC2013 competition [17].…”

Section: Experimental Analysismentioning

confidence: 99%

Depression Detection Based on Deep Distribution Learning

Melo

Granger

Hadid

2019

2019 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Major depressive disorder is among the most common and harmful mental health problems. Several deep learning architectures have been proposed for video-based detection of depression based on the facial expressions of subjects. To predict the depression level, these architectures are often modeled for regression with Euclidean loss. Consequently, they do not leverage the data distribution, nor explore the ordinal relationship between facial images and depression levels, and have limited robustness to noisy and uncertain labeling. This paper introduces a deep learning architecture for accurately predicting depression levels through distribution learning. It relies on a new expectation loss function that allows to estimate the underlying data distribution over depression levels, where expected values of the distribution are optimized to approach the ground-truth levels. The proposed approach can produce accurate predictions of depression levels even under label uncertainty. Extensive experiments on the AVEC2013 and AVEC2014 datasets indicate that the proposed architecture represents an effective approach that can outperform state-of-the-art techniques.

show abstract

“…In their experiments, PHOG has shown better results than STIPs. In our previous work [185], the temporal dynamic is captured by the LPQ-TOP features from facial region sub-volumes. Then a behavior pattern dictionary is learned through sparse coding schemes.…”

Section: Previous Workmentioning

confidence: 99%

“…Chapter 7. Computational Depression Diagnosis Analysis using Deep Learning Approach 129 Methods RMSE MAE Baseline [178] 13.61 10.88 team-australia [168] 10.45 N/A Uni-Ulm [167] 11.19 9.14 Wen [185] 10 tively. It can be seen from the table that, when joint tuning is applied, the MAE and RMSE obtained are 7.58 and 9.82 respectively on AVEC2013 database.…”

Section: Overall Performance By Fusing the Individual Modelsmentioning

confidence: 99%

Multi-Modality Human Action Recognition

Zhu¹

Self Cite

View full text Add to dashboard Cite

Human action recognition is very useful in many applications in various areas, e.g. video surveillance, HCI (Human computer interaction), video retrieval, gaming and security. Recently, human action recognition becomes an active research topic in computer vision and pattern recognition. A number of action recognition approaches have been proposed. However, most of the approaches are designed on the RGB images sequences, where the action data was collected by RGB/intensity camera. Thus the recognition performance is usually related to various occlusion, background, and lighting conditions of the image sequences. If more information can be provided along with the image sequences, more data sources other than the RGB video can be utilized, human actions could be better represented and recognized by the designed computer vision system. It is never easy to finish the Ph.D. study and write this dissertation. It would not have been possible to finish without the help of so many people in many ways. I would like to express my great gratitude to my supervisor Dr. Guodong Guo, for his guidance and support not only for this dissertation but throughout the time of my whole Ph.D. study. Thinking back the time 6 yeas ago, when I firstly came to USA pursuing my graduate degree in computer science, it is Dr. Guodong Guo, whose expertise, understanding, generous guidance, suggestions, valuable comments and support made it possible for me to work on such a exciting topic and complete my Ph.D. study. It was a great pleasure working with him.

show abstract

Automated Depression Diagnosis Based on Facial Dynamic Analysis and Sparse Coding

Cited by 106 publications

References 31 publications

Combining Global and Local Convolutional 3D Networks for Detecting Depression from Facial Expressions

Combining Global and Local Convolutional 3D Networks for Detecting Depression from Facial Expressions

Depression Detection Based on Deep Distribution Learning

Multi-Modality Human Action Recognition

Contact Info

Product

Resources

About