Pith. sign in

REVIEW 4 major objections 5 minor 51 references

On the Validity of Head Motion Patterns as Generalisable Depression Biomarkers

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Head-motion 'kinemes' learned from healthy subjects transfer across three Western depression datasets, matching or beating facial cues.

desk verdict A useful extension with a clean feature-construction story, but the regression generalisability claim rests on an unvalidated label conversion and possibly leaked cross-validation; the classification half is the stronger half. read the letter →

arxiv 2505.23427 v1 pith:3YKONF3O submitted 2025-05-29 cs.LG

Monika Gahalawat , Maneesh Bilalpur , Raul Fernandez Rojas , Jeffrey F. Cohn , Roland Goecke , Ramanathan Subramanian This is my paper
classification cs.LG
keywords kinemesheadmotiondepressionseverityestimationbinaryclassificationcross-corpusgeneralisabilityreconstructionerrorfeaturesNMF-GMMclusteringnonverbalbiomarkers
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that elementary head-motion units called kinemes are efficient and generalisable biomarkers for depression severity. Kinemes are learned only from head movements of healthy or low-depressed people, then used to reconstruct the movements of both low- and high-depressed groups. Across three culturally distinct Western datasets (German AVEC2013, Australian Blackdog, American Pitt), kineme-based features match or beat raw head-pose features for classifying low versus severe depression and outperform other visual cues for severity regression. The strongest single result is an MAE of 5.68 on the AVEC2013 test set, second only to a facial-keypoint method. A reliable head-motion-only marker would matter because it is cheap to extract from ordinary video and reveals less about identity than facial appearance.

What carries the argument

The kineme dictionary is a set of k = 16 prototypical five-second head-pose segments learned via non-negative matrix factorisation (to obtain a basis and coefficient matrix) followed by Gaussian mixture model clustering in the coefficient space, built only from healthy or low-depressed subjects. Any subject's head motion is then expressed as a sequence of nearest kinemes, and the signed difference between original and reconstructed pose, summarised as minimum, maximum, range, mean, median, standard deviation, skewness, and kurtosis over each of pitch, yaw, and roll, forms the feature vector. This reconstruction-error design is what forces the model to encode departures from normal head motion, and the control-only learning is what makes the features a candidate biomarker rather than a within-corpus artifact.

What would settle it

Re-run the k-fold and separate train-test experiments with strictly subject-exclusive splits, ensuring all videos of one participant appear in only one fold or one set. If the cross-dataset F1 and MAE gaps shrink toward chance, or if the AVEC-derived kineme advantage disappears, the claimed generalisability is an artifact of participant leakage across splits.

Watch

Extended reading notes

Core claim

The paper's central discovery is that the mismatch between observed head-pose segments and a small learned dictionary of typical head-motion patterns (kinemes) carries depression-severity information that transfers across datasets. Kinemes are discovered by NMF decomposition and GMM clustering of overlapping five-second yaw, pitch, and roll segments taken only from healthy controls; eight statistical descriptors of the per-axis reconstruction error yield a 24-dimensional feature vector. In both k-fold cross-validation and separate train-test reuse experiments, kinemes derived from AVEC generalise best, and the AVEC-derived set achieves an F1 score of 0.77 across all three datasets while outperforming raw head-pose features for classification and facial-feature CNN baselines for regression. On the AVEC2013 test set, XGBoost regression on these features gives an MAE of 5.68 and RMSE of 7.57.

Load-bearing premise

The load-bearing premise is that the train and test splits never put the same person in both sets; the paper does not state that cross-validation or train-test splits are subject-disjoint, and both AVEC and Pitt contain multiple videos per participant.

Editorial extensions

If this is right

  • Head-pose-only pipelines are viable for depression screening, avoiding dependency on facial appearance features and their privacy costs.
  • Kineme features transfer across recording setups (self-administered tasks versus clinical interviews) and across language and cultural groups within Western data.
  • Kinemes discovered from self-directed task recordings such as AVEC can serve as a shared dictionary for future cross-corpus depression studies.
  • Regression to a common severity scale (QIDS-equivalent) works from head motion alone, with error rates competitive with facial-feature deep learning models.
  • Because the final feature set is only 24 numbers, the pipeline is lightweight enough for near-real-time screening with classical machine learning classifiers.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because AVEC contains 150 videos from 82 participants and Pitt contains 148 videos from 49 participants, the claimed generalisability depends on train-test splits being subject-disjoint; the paper does not state this, so a subject-exclusive resplit is the first test I would run.
  • The Blackdog labels (healthy controls versus depressed patients) differ in meaning from the AVEC and Pitt labels (low versus severe depression), which likely explains the weaker transfer onto Blackdog; calibrating kineme features to symptom-severity scales rather than diagnostic class could make cross-dataset transfer more clinically meaningful.
  • The finding that AVEC-derived kinemes generalise better than interview-derived kinemes suggests task context matters more than culture, a hypothesis that could be tested directly by learning kinemes from two interview datasets and evaluating on a third.
  • The reconstruction-error framing could yield an interpretable clinical readout, such as per-axis error profiles showing which head-motion patterns deviate most, although the paper does not pursue that visualisation.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This manuscript examines whether head-motion-based "kinemes"—elementary motion atoms discovered via NMF and GMM clustering—can serve as generalisable depression biomarkers across three Western datasets (German AVEC2013, American Pitt, Australian Blackdog). Kinemes are learned only from low-depressed/healthy subjects, and all videos are represented by statistical descriptors of the reconstruction error between actual and kineme-reconstructed head pose segments. The authors evaluate binary classification and severity regression using k-fold cross-validation and cross-dataset train-test reuse, and claim that (1) kinemes are efficient biomarkers (second-best MAE on AVEC2013) and (2) kineme features generalise better than raw head pose descriptors for classification and than other visual cues for regression.

Significance. If the results are valid, this is a practically useful contribution: it suggests that a compact, interpretable head-motion representation can rival much more complex deep facial models on cross-cultural depression severity estimation, and it provides one of the few three-dataset generalisability studies in the field. The kineme construction is non-circular and transparent, and the use of three public datasets is a clear strength. However, unresolved issues around label scaling, subject-disjoint evaluation, and post-hoc model selection currently prevent the reported effect sizes from being taken at face value.

major comments (4)
  1. [Section 4, Tables 4-5 and 7-8] The conversion of AVEC BDI and Pitt HRSD labels to QIDS-SR via an IDS/QIDS conversion table is not validated and appears to be applied inconsistently. If Table 4's AVEC CV MAE of 3.16 is on the converted QIDS-SR scale (range 0-27) while Table 8's AVEC test MAE of 5.68 is compared against prior works that evaluate on the original BDI scale (range 0-63), the 'second best MAE' claim is not supported because the metrics are not commensurable. The authors should either (i) validate the cross-instrument mapping against clinical concordance data, (ii) rerun all regression experiments on the original instrument for each dataset, or (iii) restrict the claim to comparisons within the same label scale.
  2. [Section 5.3, Tables 2 and 4] The k-fold cross-validation splits videos, not participants, and the manuscript does not state that folds are subject-disjoint. Since AVEC contains 150 videos from 82 participants and Pitt contains 148 videos from 49 participants, the same person can appear in both training and test folds, which inflates the reported within-dataset F1/MAE values (e.g., AVEC F1 = 0.93 in Table 2) and weakens the claim that kinemes are efficient biomarkers for same-distribution data. The cross-dataset separate train-test experiments (Tables 3 and 5) are not affected by this leakage, but the paper should either report subject-disjoint folds or explicitly qualify the within-dataset results.
  3. [Sections 5.3-5.4, Tables 2-5] The model and chunk-size selection is post hoc: Section 5.4 states 'we report the best results obtained' among Random Forest, SVM, and XGB, and Section 6 states that results are compiled with a chunk size of 60s after trying 60/75/90/120s. Without a predetermined held-out validation protocol, these choices can overfit the test data and bias the comparisons in Tables 6-8. Please describe the selection procedure (e.g., nested cross-validation or a separate validation fold) and separate it from the final evaluation on the test set.
  4. [Section 6.1, Table 6] The key classification comparison with prior work is not metric-aligned: the authors report accuracy, F1, precision, and recall for their kineme method, but the baseline column from Alghowinem et al. [30] only provides 'Average Recall.' The claim that kineme features are more generalisable than raw head pose features for classification therefore rests on comparing F1 or accuracy with average recall, which is not a like-for-like comparison. Please re-evaluate the prior method on the same metrics (or obtain the same metrics from the original implementation) or explicitly justify why a cross-metric comparison is meaningful.
minor comments (5)
  1. [Section 5.3] '5-repetitions of 10-fold cross-validation (50 runs)' is imprecise; five repeats of a 10-fold CV yields 50 fold evaluations, not 50 independent runs.
  2. [Section 4 and throughout] The dataset name appears as 'A VEC' in the first paragraph of Section 4 and in several tables, but as 'AVEC' elsewhere; please standardise the spelling.
  3. [Section 6, third bullet] The F1 value 0.77 attributed to Table 3 for the AVEC-conf condition actually appears in Table 2 (row 'All three'); the cross-referencing should be corrected.
  4. [Reference [34]] There is a typo in the title: 'Comparison of Beck depression inventories-ia and-ii' should read 'Comparison of Beck Depression Inventories-IA and -II.'
  5. [Tables 5 and 7 captions] The captions state 'Lowest RMSE/MAE values per row are denoted in bold,' but no bold formatting appears in the provided text; please ensure the final typeset version matches the captions.

Circularity Check

1 steps flagged · score 6.0 of 10

Same-corpus CV predictions for the low-depressed class reduce to the kineme dictionary fit.

  1. fitted input called prediction [Section 5.2 (Kineme Discovery) and Section 5.3 (Implementation Details), Eqs. (1)-(4)]
    "To examine generalisability of learned kineme patterns, we implemented three separate configurations to discover kinemes from the low-depressed/healthy class. For the AVEC configuration (AVEC-conf), we initially discovered 16 kinemes using the head motion data acquired from low-depressed videos."

    Section 5.2 learns 16 kinemes from all low-depressed videos of a configuration; Section 5.3 then applies 5x10-fold CV to features computed from those same videos, and the paper does not state that kinemes are re-discovered inside each training fold. Eq. (1) minimizes ||H-BC||_F over H assembled from those low-depressed segments, and Eq. (4) defines the feature d(i)=h(i)-h~(i) used for classification and regression. Hence a low-depressed test segment's reconstruction error is minimized by construction, since it helped choose B and the GMM centers. The classifier's ability to separate low-depressed from severely depressed in Tables 2 and 4 is therefore partly a replay of the fit rather than an independent prediction.

full rationale

The feature-construction chain itself is not circular: kinemes are learned from head pose of low-depressed/healthy subjects only, and outcome labels enter only in the final classifier or regressor. The main circularity is in the same-corpus evaluation protocol. Section 5.2 describes discovering kinemes once per configuration from all low-depressed videos; Section 5.3 then runs k-fold CV on the resulting features without stating that the dictionary is re-fit per training fold. Under that protocol, test-fold low-depressed videos are included in the NMF/GMM fit (Eq. 1), so their reconstruction-error features (Eq. 4) are small by construction. This inflates the same-dataset rows of Tables 2 and 4 and the corresponding 'kinemes vs raw head pose' comparisons in Table 6. The cross-dataset (separate train-test) experiments, which are the core of the generalisability claim, do not share this flaw because the target dataset is never used for kineme discovery; those results give the central claim independent content. The unvalidated BDI/HRSD-to-QIDS conversion and the lack of explicit subject-disjoint splits are validity concerns but not circularity. The moderate score reflects the partial, evaluation-protocol circularity rather than a definitional equivalence of the whole method.

Assumptions & free parameters 4 free parameters · 5 assumptions · 1 invented entities

The central claim rests on a small set of engineering choices: the number of kinemes, the NMF rank, and the chunk size are set by hand or inherited from prior work, and the regression task additionally depends on cross-scale label conversion. No new physical entities are proposed; kinemes are a data-driven representation rather than an independently evidenced construct.

free parameters (4)
  • Number of kinemes k = 16
    Set to 16 for all configurations following prior work [20]; no sensitivity analysis is reported, so the value is chosen rather than derived.
  • NMF subspace rank q = not reported
    Equation (1) allows any q <= min(m,n), but the rank actually used for GMM clustering is not specified, which is a free choice with unknown effect on kineme quality.
  • Chunk size n_c = 60 s used for final results (75, 90, 120 s also tested)
    The paper reports video-level results for n_c=60s after stating that chunk sizes 60, 75, 90, and 120 were processed; if 60s was selected after observing test performance, it is a fitted choice.
  • Classifier and regressor hyperparameters = not reported
    For separate train-test runs, parameters were set to those giving highest training accuracy, and for AVEC test runs parameters were tuned on the validation set; no grid or regularization specifics are given.
assumptions (5)
  • domain assumption Head pose angles can be shifted to [0,360] degrees without losing information needed for NMF non-negativity.
    Section 3.1 states angles are maintained in [0,360] to ensure non-negativity, which may distort circular angle continuity.
  • domain assumption OpenFace provides accurate 3D head pose angles from cropped video frames across all three datasets.
    Section 3.1 and 5.1 rely on OpenFace pose extraction after face cropping, with no manual validation or reported pose error.
  • domain assumption BDI, HRSD, and QIDS-SR severity scores can be mapped to a common QIDS-SR scale using published conversion tables.
    Section 4 converts labels for AVEC and Pitt to QIDS-SR with the IDS and QIDS conversion table; the conversion is not validated on these datasets and directly feeds the regression targets.
  • domain assumption Video-level chunks are treated as independent samples even when the same subject appears in multiple videos.
    Section 5.3 describes video-level analysis with repeated labels per chunk; AVEC has 150 videos from 82 participants and Pitt has 148 videos from 49 participants, so splits that are not subject-disjoint would leak identity.
  • domain assumption Reconstruction error between raw head pose and kineme cluster centers is a useful depression severity signal.
    This is the core feature premise, inherited from [20]; the paper does not independently motivate why this particular reconstruction error should correlate with depression.
invented entities (1)
  • Kinemes (elementary head motion units)
    purpose: Discrete prototypical head-motion segments used to compute reconstruction-error features for depression classification and regression.
    Introduced in the authors' prior work [20] and learned from data via NMF and GMM; no external, physiological, or clinical validation is provided in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On the Validity of Head Motion Patterns as Generalisable Depression Biomarkers." pith.science (2026). https://pith.science/paper/3YKONF3O

@misc{pith2026250523427,
  author       = {Pith},
  title        = {Pith review of: On the Validity of Head Motion Patterns as Generalisable Depression Biomarkers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3YKONF3O}},
  note         = {Machine review of arXiv:2505.23427}
}
read the original abstract

Depression is a debilitating mood disorder negatively impacting millions worldwide. While researchers have explored multiple verbal and non-verbal behavioural cues for automated depression assessment, head motion has received little attention thus far. Further, the common practice of validating machine learning models via a single dataset can limit model generalisability. This work examines the effectiveness and generalisability of models utilising elementary head motion units, termed kinemes, for depression severity estimation. Specifically, we consider three depression datasets from different western cultures (German: AVEC2013, Australian: Blackdog and American: Pitt datasets) with varied contextual and recording settings to investigate the generalisability of the derived kineme patterns via two methods: (i) k-fold cross-validation over individual/multiple datasets, and (ii) model reuse on other datasets. Evaluating classification and regression performance with classical machine learning methods, our results show that: (1) head motion patterns are efficient biomarkers for estimating depression severity, achieving highly competitive performance for both classification and regression tasks on a variety of datasets, including achieving the second best Mean Absolute Error (MAE) on the AVEC2013 dataset, and (2) kineme-based features are more generalisable than (a) raw head motion descriptors for binary severity classification, and (b) other visual behavioural cues for severity estimation (regression).

Figures

Figures reproduced from arXiv: 2505.23427 by the authors.

Figure 1
Figure 1. Overview: Kineme patterns are solely learnt from control subjects, and the reconstruction error is computed between the actual vs. reconstructed head-motion segments for both the control and depressed classes. Statistical descriptors over the yaw, pitch and roll angular dimensions (8 × 3 features) are utilised for depression severity estimation. [20] tude, velocity and acceleration of 3D head pose angles from all cl… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

51 extracted references · 46 canonical work pages

  1. [20]

    Explainable depression detection via head motion patterns,

    M. Gahalawat, R. Fernandez Rojas, T. Guha, R. Subramanian, and R. Goecke, “Explainable depression detection via head motion patterns,” inProceedings of the 25th International Conference on Multimodal Interaction, 2023, pp. 261–270

    work page 2023

  2. [30]

    Cross-cultural detection of depression from nonverbal behaviour,

    S. Alghowinem, R. Goecke, J. F. Cohn, M. Wagner, G. Parker, and M. Breakspear, “Cross-cultural detection of depression from nonverbal behaviour,” in2015 11th IEEE International conference and workshops on automatic face and gesture recognition (FG), vol. 1. IEEE, 2015, pp. 1–8

    work page 2015

  3. [1]

    Global health data exchange (ghdx),

    I. of Health Metrics and Evaluation, “Global health data exchange (ghdx),” 2021

    work page 2021

  4. [2]

    Suicidal ideation in a random community sample: attributable risk due to depression and psychosocial and traumatic events,

    R. D. Goldney, D. Wilson, E. D. Grande, L. J. Fisher, and A. C. McFarlane, “Suicidal ideation in a random community sample: attributable risk due to depression and psychosocial and traumatic events,”Australian & New Zealand Journal of Psychiatry, vol. 34, no. 1, pp. 98–106, 2000

    work page 2000

  5. [3]

    The economic burden of adults with major depressive disorder in the united states (2005 and 2010),

    P . E. Greenberg, A.-A. Fournier, T. Sisitsky, C. T. Pike, and R. C. Kessler, “The economic burden of adults with major depressive disorder in the united states (2005 and 2010),”The Journal of clinical psychiatry, vol. 76, no. 2, p. 5356, 2015

    work page 2005

  6. [4]

    Indirect costs of depression and other mental and behavioural disorders for australia from 2015 to 2030,

    D. Schofield, M. Cunich, R. Shrestha, R. Tanton, L. Veerman, S. Kelly, and M. Passey, “Indirect costs of depression and other mental and behavioural disorders for australia from 2015 to 2030,” BJPsych open, vol. 5, no. 3, p. e40, 2019

    work page 2015

  7. [5]

    An exploration of residents’ implicit biases towards depression—a pilot study,

    K. Crapanzano, D. Fisher, R. Hammarlund, E. P . Hsieh, and W. May, “An exploration of residents’ implicit biases towards depression—a pilot study,”Journal of General Internal Medicine, vol. 33, pp. 2065–2069, 2018

    work page 2018

  8. [6]

    Multimodal assessment of depression from behavioral signals,

    J. F. Cohn, N. Cummins, J. Epps, R. Goecke, J. Joshi, and S. Scherer, “Multimodal assessment of depression from behavioral signals,” The Handbook of Multimodal-Multisensor Interfaces: Signal Processing, Architectures, and Detection of Emotion and Cognition-Volume 2, pp. 375–417, 2018

    work page 2018

Show all 51 references
  1. [7]

    Automatic Assessment of De- pression Based on Visual Cues: A Systematic Review,

    A. Pampouchidou, P . G. Simos, K. Marias, F. Meriaudeau, F. Yang, M. Pediaditis, and M. Tsiknakis, “Automatic Assessment of De- pression Based on Visual Cues: A Systematic Review,”IEEE T rans- actions on Affective Computing, vol. 10, no. 4, pp. 445–470, 2019

    2019

  2. [8]

    Processing of facial emo- tion expression in major depression: a review,

    C. Bourke, K. Douglas, and R. Porter, “Processing of facial emo- tion expression in major depression: a review,”Australian & New Zealand Journal of Psychiatry, vol. 44, no. 8, pp. 681–696, 2010

    2010

  3. [9]

    Multimodal depression detection: fusion analysis of paralinguistic, head pose and eye gaze behav- iors,

    S. Alghowinem, R. Goecke, M. Wagner, J. Epps, M. Hyett, G. Parker, and M. Breakspear, “Multimodal depression detection: fusion analysis of paralinguistic, head pose and eye gaze behav- iors,”IEEE T ransactions on Affective Computing, vol. 9, no. 4, pp. 478–490, 2016

    2016

  4. [10]

    Relative body parts JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2024 8 movement for automatic depression analysis,

    J. Joshi, A. Dhall, R. Goecke, and J. F. Cohn, “Relative body parts JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2024 8 movement for automatic depression analysis,” in2013 Humaine association conference on affective computing and intelligent interaction. IEEE, 2013, pp...

    2024

  5. [11]

    Head pose and movement analysis as an indicator of depression,

    S. Alghowinem, R. Goecke, M. Wagner, G. Parkerx, and M. Break- spear, “Head pose and movement analysis as an indicator of depression,” in2013 Humaine Association Conference on Affective Computing and Intelligent Interaction. IEEE, 2013, pp. 283–288

    2013

  6. [12]

    An in- vestigation of depressed speech detection: Features and normal- ization,

    N. Cummins, J. Epps, M. Breakspear, and R. Goecke, “An in- vestigation of depressed speech detection: Features and normal- ization,” inTwelfth Annual Conference of the International Speech Communication Association, 2011

    2011

  7. [13]

    Deep multi-modal network based automated depression severity estimation,

    M. A. Uddin, J. B. Joolee, and K.-A. Sohn, “Deep multi-modal network based automated depression severity estimation,”IEEE transactions on affective computing, vol. 14, no. 3, pp. 2153–2167, 2022

    2022

  8. [14]

    A multimodal fusion model with multi-level attention mechanism for depression detection,

    M. Fang, S. Peng, Y. Liang, C.-C. Hung, and S. Liu, “A multimodal fusion model with multi-level attention mechanism for depression detection,”Biomedical Signal Processing and Control, vol. 82, p. 104561, 2023

    2023

  9. [15]

    Avec 2013: the continuous audio/visual emotion and depression recognition challenge,

    M. Valstar, B. Schuller, K. Smith, F. Eyben, B. Jiang, S. Bilakhia, S. Schnieder, R. Cowie, and M. Pantic, “Avec 2013: the continuous audio/visual emotion and depression recognition challenge,” in Proceedings of the 3rd ACM international workshop on Audio/visual emotion challe...

    2013

  10. [16]

    Detecting depression sever- ity from vocal prosody,

    Y. Yang, C. Fairbairn, and J. F. Cohn, “Detecting depression sever- ity from vocal prosody,”IEEE transactions on affective computing, vol. 4, no. 2, pp. 142–150, 2012

    2012

  11. [17]

    The distress analysis interview corpus of human and computer interviews

    J. Gratch, R. Artstein, G. M. Lucas, G. Stratou, S. Scherer, A. Nazar- ian, R. Wood, J. Boberg, D. DeVault, S. Marsellaet al., “The distress analysis interview corpus of human and computer interviews.” in LREC. Reykjavik, 2014, pp. 3123–3128

    2014

  12. [18]

    Cultural dif- ferences in the development and characteristics of depression

    J. Gabriella, E. N ´ora, P . Dorottya, and G. X ´enia, “Cultural dif- ferences in the development and characteristics of depression.” Neuropsychopharmacologia Hungarica, 2012

    2012

  13. [19]

    Cultural aspects of depressive experience and disorders,

    A. J. Marsella, “Cultural aspects of depressive experience and disorders,”Online readings in psychology and culture, vol. 10, no. 2, p. 4, 2003

    2003

  14. [21]

    Nonverbal cues for depression

    P . Waxer, “Nonverbal cues for depression.”Journal of Abnormal Psychology, vol. 83, no. 3, p. 319, 1974

    1974

  15. [22]

    An ethological description of depression,

    J. Pedersen, J. Schelde, E. Hannibal, K. Behnke, B. Nielsen, and M. Hertz, “An ethological description of depression,”Acta psychi- atrica scandinavica, vol. 78, no. 3, pp. 320–330, 1988

    1988

  16. [23]

    The ethological approach to the assessment of depressive disorders,

    L. Fossi, C. Faravelli, and M. Paoli, “The ethological approach to the assessment of depressive disorders,”The Journal of nervous and mental disease, vol. 172, no. 6, pp. 332–341, 1984

    1984

  17. [24]

    Relationship between depression and movement quality in normal young adults,

    S. Mitsue and T. Yamamoto, “Relationship between depression and movement quality in normal young adults,”Journal of Physical Therapy Science, vol. 31, no. 10, pp. 819–822, 2019

    2019

  18. [25]

    Evaluating the severity of depressive symptoms using upper body motion captured by rgb-depth sensors and machine learning in a clinical interview setting: a preliminary study,

    T. Horigome, B. Sumali, M. Kitazawa, M. Yoshimura, K.-c. Liang, Y. Tazawa, T. Fujita, M. Mimura, and T. Kishimoto, “Evaluating the severity of depressive symptoms using upper body motion captured by rgb-depth sensors and machine learning in a clinical interview setting: a prel...

    2020

  19. [26]

    Investigation of speech landmark patterns for depression detection,

    Z. Huang, J. Epps, and D. Joachim, “Investigation of speech landmark patterns for depression detection,”IEEE T ransactions on Affective Computing, vol. 13, no. 2, pp. 666–679, 2019

    2019

  20. [27]

    Can body expressions contribute to automatic depression analysis?

    J. Joshi, R. Goecke, G. Parker, and M. Breakspear, “Can body expressions contribute to automatic depression analysis?” in2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG). IEEE, 2013, pp. 1–7

    2013

  21. [28]

    Detecting depression severity by interpretable representations of motion dynamics,

    A. Kacem, Z. Hammal, M. Daoudi, and J. Cohn, “Detecting depression severity by interpretable representations of motion dynamics,” in2018 13th ieee international conference on automatic face & gesture recognition (fg 2018). IEEE, 2018, pp. 739–745

    2018

  22. [29]

    Interpretation of depression detection models via feature selec- tion methods,

    S. M. Alghowinem, T. Gedeon, R. Goecke, J. Cohn, and G. Parker, “Interpretation of depression detection models via feature selec- tion methods,”IEEE transactions on affective computing, 2020

    2020

  23. [31]

    CNN depression severity level estimation from upper body vs. face-only images,

    D. Ahmad, R. Goecke, and J. Ireland, “CNN depression severity level estimation from upper body vs. face-only images,” inIn- ternational Conference on Pattern Recognition. Springer, 2021, pp. 744–758

    2021

  24. [32]

    OpenFace: An open source facial behavior analysis toolkit,

    T. Baltru ˇsaitis, P . Robinson, and L.-P . Morency, “OpenFace: An open source facial behavior analysis toolkit,” in2016 IEEE Winter Conference on Applications of Computer Vision, 2016, pp. 1–10

    2016

  25. [33]

    On the role of head motion in affective expression,

    A. Samanta and T. Guha, “On the role of head motion in affective expression,” inIEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2017, pp. 2886–2890

    2017

  26. [34]

    Comparison of beck depression inventories-ia and-ii in psychiatric outpatients,

    A. T. Beck, R. A. Steer, R. Ball, and W. F. Ranieri, “Comparison of beck depression inventories-ia and-ii in psychiatric outpatients,” Journal of personality assessment, vol. 67, no. 3, pp. 588–597, 1996

    1996

  27. [35]

    Automated depression diagnosis based on facial dynamic analysis and sparse coding,

    L. Wen, X. Li, G. Guo, and Y. Zhu, “Automated depression diagnosis based on facial dynamic analysis and sparse coding,” IEEE T ransactions on Information Forensics and Security, vol. 10, no. 7, pp. 1432–1441, 2015

    2015

  28. [36]

    Automated depression diagnosis based on deep networks to encode facial appearance and dynamics,

    Y. Zhu, Y. Shang, Z. Shao, and G. Guo, “Automated depression diagnosis based on deep networks to encode facial appearance and dynamics,”IEEE T ransactions on Affective Computing, vol. 9, no. 4, pp. 578–584, 2017

    2017

  29. [37]

    Visually interpretable representation learning for depression recognition from facial images,

    X. Zhou, K. Jin, Y. Shang, and G. Guo, “Visually interpretable representation learning for depression recognition from facial images,”IEEE transactions on affective computing, vol. 11, no. 3, pp. 542–552, 2018

    2018

  30. [38]

    Depression severity prediction from facial expression based on the drr depressionnet network,

    X. Li, W. Guo, and H. Yang, “Depression severity prediction from facial expression based on the drr depressionnet network,” in 2020 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE, 2020, pp. 2757–2764

    2020

  31. [39]

    A deep multiscale spatiotemporal network for assessing depression from facial dy- namics,

    W. C. De Melo, E. Granger, and A. Hadid, “A deep multiscale spatiotemporal network for assessing depression from facial dy- namics,”IEEE transactions on affective computing, vol. 13, no. 3, pp. 1581–1592, 2020

    2020

  32. [40]

    Automatic depression recognition using cnn with attention mechanism from videos,

    L. He, J. C.-W. Chan, and Z. Wang, “Automatic depression recognition using cnn with attention mechanism from videos,” Neurocomputing, vol. 422, pp. 165–175, 2021

    2021

  33. [41]

    Estimating depression severity from long-sequence face videos via an ensemble global diverse convolutional model,

    G. Bargshady and R. Goecke, “Estimating depression severity from long-sequence face videos via an ensemble global diverse convolutional model,” in2023 International Conference on Digital Image Computing: T echniques and Applications (DICTA). IEEE, 2023, pp. 296–303

    2023

  34. [42]

    Inte- grating deep facial priors into landmarks for privacy preserving multimodal depression recognition,

    Y. Pan, Y. Shang, Z. Shao, T. Liu, G. Guo, and H. Ding, “Inte- grating deep facial priors into landmarks for privacy preserving multimodal depression recognition,”IEEE T ransactions on Affective Computing, 2023

    2023

  35. [43]

    Pra-net: Part-and-relation attention network for depression recognition from facial expression,

    Z. Liu, X. Yuan, Y. Li, Z. Shangguan, L. Zhou, and B. Hu, “Pra-net: Part-and-relation attention network for depression recognition from facial expression,”Computers in Biology and Medicine, vol. 157, p. 106589, 2023

    2023

  36. [44]

    Two-stage temporal modelling framework for video-based depression recog- nition using graph representation,

    J. Xu, H. Gunes, K. Kusumam, M. Valstar, and S. Song, “Two-stage temporal modelling framework for video-based depression recog- nition using graph representation,”IEEE T ransactions on Affective Computing, 2024

    2024

  37. [45]

    Spatial–temporal attention network for depression recognition from facial videos,

    Y. Pan, Y. Shang, T. Liu, Z. Shao, G. Guo, H. Ding, and Q. Hu, “Spatial–temporal attention network for depression recognition from facial videos,”Expert Systems with Applications, vol. 237, p. 121410, 2024

    2024

  38. [46]

    Depressionmlp: A multi-layer perceptron architecture for automatic depression level prediction via facial keypoints and action units,

    M. Niu, Y. Li, J. Tao, X. Zhou, and B. W. Schuller, “Depressionmlp: A multi-layer perceptron architecture for automatic depression level prediction via facial keypoints and action units,”IEEE T rans- actions on Circuits and Systems for Video T echnology, 2024

    2024

  39. [47]

    Spectral repre- sentation of behaviour primitives for depression analysis,

    S. Song, S. Jaiswal, L. Shen, and M. Valstar, “Spectral repre- sentation of behaviour primitives for depression analysis,”IEEE T ransactions on Affective Computing, vol. 13, no. 2, pp. 829–844, 2020

    2020

  40. [48]

    Imagenet classifica- tion with deep convolutional neural networks,

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classifica- tion with deep convolutional neural networks,”Advances in neural information processing systems, vol. 25, 2012

    2012

  41. [49]

    Very deep convolutional networks for large-scale image recognition,

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,”arXiv preprint arXiv:1409.1556, 2014

    2014 arXiv

  42. [50]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2016, pp. 770–778

    2016

  43. [51]

    Densely connected convolutional networks,

    G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, “Densely connected convolutional networks,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 4700–4708

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.