Pith. sign in

REVIEW 4 major objections 5 minor 41 references

EmoNeXt: an Adapted ConvNeXt for Facial Emotion Recognition

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read EmoNeXt claims 76.12% on FER2013, a new state of the art

desk verdict Clean architecture study with a SOTA claim that's within noise; worth a referee but not a citation as established. read the letter →

arxiv 2501.08199 v1 pith:O6KW2DTP submitted 2025-01-14 cs.CV cs.AI

Yassine El Boudouri , Amine Bohi This is my paper · ORCID
classification cs.CVcs.AI
keywords FacialexpressionrecognitionFER2013ConvNeXtSpatialtransformernetworkSqueeze-and-excitationSelf-attentionregularizationEmotionclassificationDeeplearning
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 claims that a convolutional architecture built from ConvNeXt, with a Spatial Transformer Network in front, Squeeze-and-Excitation blocks between stages, and a self-attention regularization term in the loss, sets a new state of the art on the FER2013 facial emotion recognition benchmark. The largest variant, EmoNeXt-XLarge, reports 76.12% test accuracy, edging past the previous best reported by Segmentation VGG-19 at 75.97%. The same recipe improves all five model sizes over their plain ConvNeXt counterparts, so the authors argue that the gains come from the added components rather than from raw scale. A sympathetic reader would care because FER2013 is a widely used, unbalanced, in-the-wild benchmark, and any reproducible improvement on it is useful for applications such as human-machine interaction, healthcare, and driver monitoring.

What carries the argument

The central object is the EmoNeXt architecture: a ConvNeXt backbone preceded by a Spatial Transformer Network and followed at each stage by a Squeeze-and-Excitation block, trained with cross-entropy plus a self-attention regularization term. The STN is a differentiable module that predicts a spatial transformation and resamples the input, letting the network align faces despite scale, rotation, and translation variation. The SE blocks squeeze each channel to a scalar with global average pooling and excite channel-wise weights through fully connected layers, adaptively emphasizing informative channels. The self-attention regularization term is the variance of the dot-product attention weights, added to the loss with a trade-off hyperparameter $\lambda$; it drives the attention distribution toward uniformity so features are used more evenly. Together these components are what the paper credits for the accuracy gains over plain ConvNeXt and earlier FER models.

What would settle it

Re-run EmoNeXt-XLarge and the strongest Table II baselines, especially Segmentation VGG-19, on the same FER2013 test images using identical preprocessing, augmentation, pretrained weights, and evaluation code. If any baseline then matches or exceeds 76.12%, or if EmoNeXt-XLarge drops below 75.97% under the shared protocol, the paper's central superiority claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that adding three mechanisms to ConvNeXt yields a better facial emotion classifier: an STN that learns affine-style spatial transformations to align faces, SE blocks that recalibrate channel-wise feature responses after each ConvNeXt stage, and a self-attention regularization term. The regularization term is $L_{\mathrm{SA}} = \frac{1}{N}\sum_{i=1}^{N}(W_i - \bar{W})^2$, the variance of dot-product self-attention weights, added to cross-entropy as $L_{\mathrm{final}} = L_{\mathrm{CE}} + \lambda L_{\mathrm{SA}}$. Minimizing this variance pushes attention weights toward their mean, encouraging balanced, compact feature vectors. On the FER2013 test set, EmoNeXt-XLarge reaches 76.12% accuracy, surpassing the 75.97% reported for Segmentation VGG-19, and every EmoNeXt size outperforms the corresponding ConvNeXt size.

Load-bearing premise

The load-bearing premise is that the published accuracy numbers for the earlier models in Table II are accurate and directly comparable to EmoNeXt's, even though they come from different groups with different training protocols; the claimed superiority rests on a mere 0.15 percentage-point margin over the previous best.

Editorial extensions

If this is right

  • EmoNeXt-XLarge claims 76.12% on FER2013, a 0.15 percentage-point gain over Segmentation VGG-19's 75.97%, making it the paper's new single-model state of the art.
  • The same three additions lift every size of ConvNeXt: EmoNeXt-Tiny reaches 73.34%, Small 74.33%, Base 74.91%, Large 75.57%, and XLarge 76.12%.
  • Smaller EmoNeXt variants outperform larger plain ConvNeXt models, e.g., EmoNeXt-Small (74.33%) beats ConvNeXt-XLarge (74.15%), suggesting the architectural additions matter more than parameter count.
  • The full training recipe, including ImageNet-22k pretrained weights, AdamW, cosine decay, stochastic depth, label smoothing, EMA, and mixed precision, is what produced these numbers, so the result is tied to modern training practices, not only the architecture.

Reading between the lines

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

  • The comparison in Table II mixes accuracy numbers reported by different groups with different training protocols; a direct head-to-head re-run on the same FER2013 test split is the cleanest way to see whether the 0.15-point edge survives.
  • Because the self-attention regularization term is defined on feature-vector attention weights and added to any cross-entropy loss, it is not tied to ConvNeXt; it could be tested as a plug-in auxiliary loss for ResNet, VGG, or transformer-based FER models.
  • The reported improvements are all on FER2013. The paper's stated next step, emotion recognition for Alzheimer's patients, would require cross-dataset and population-specific validation before the architecture's practical value in clinical settings is established.
  • If the SA regularization indeed works by flattening attention distributions, one testable extension is to ablate $\lambda$ and visualize the attention maps: the claim predicts less concentrated attention and more evenly distributed feature use as $\lambda$ grows.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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. The paper proposes EmoNeXt, an adaptation of the ConvNeXt architecture for facial expression recognition on FER2013. The model adds a Spatial Transformer Network (STN), Squeeze-and-Excitation (SE) blocks after each ConvNeXt stage, and a self-attention regularization term (SA) added to the cross-entropy loss. The authors report results for five model sizes (Tiny through XLarge) and claim that EmoNeXt-XLarge reaches 76.12% accuracy, surpassing the previous best of 75.97% from Segmentation VGG-19. The manuscript includes architecture descriptions, training details, and a comparison table, plus a link to code.

Significance. If the reported accuracy is reproducible under matched experimental conditions, EmoNeXt would represent a modest but legitimate state-of-the-art result on FER2013, a widely used benchmark. The paper's strengths include a clear architectural description, the release of code, and evaluation across multiple model scales. However, the central claim rests on a single accuracy number per model with no variance estimates, and the comparison mixes results from independently published baselines while claiming a common experimental setup. The significance of the claimed improvement is therefore not established by the present evidence; the contribution is better viewed as an architecture proposal whose empirical validation needs substantial strengthening.

major comments (4)
  1. [Section IV-C and Table II] The claim of state-of-the-art performance relies on comparing EmoNeXt rows against published baseline numbers from at least six different papers ([14], [15], [16], [19], [36]-[40]), yet the Introduction states that comparisons are made "under the same experimental setup" and Section IV-C repeats this wording. Only the ConvNeXt and EmoNeXt rows appear to be generated by this work. Different pretraining, augmentation, and evaluation protocols across these baselines make the 0.15 percentage point margin over Segmentation VGG-19 non-comparable as reported. The authors should either reproduce the baselines in their own pipeline or temper the superiority claim accordingly.
  2. [Section IV-C and Table II] No error bars, multiple seeds, or statistical tests are provided for any accuracy number. With a test set of 3,589 images and accuracy near 76%, the standard error of the difference between two proportions is roughly sqrt(2 * 0.76 * 0.24 / 3589) ≈ 1.0 percentage point, so the reported margin of 0.15 percentage points is about 0.15 standard errors. The claimed advantage over Segmentation VGG-19 is thus not statistically resolvable from the reported numbers, independent of protocol comparability. The authors should report the mean and standard deviation over several runs and perform a significance test for the key comparison.
  3. [Section III-E, Eq. (3), and Section IV-B] The value of the regularization weight lambda in Eq. (3) is never reported, nor is the selection procedure. Without this value, the self-attention regularization term cannot be reproduced or evaluated. Additionally, no ablation study isolates the contributions of the STN, the SE blocks, and the SA term; since the paper's contribution is precisely these additions, an ablation is necessary to support the claim that each component helps.
  4. [Section IV-B and Conclusion] Training details are incomplete: the batch size, number of epochs, exact learning rate schedule, data augmentation hyperparameters, and STN/SE configuration are not specified, making exact reproduction difficult. Furthermore, the Conclusion states that "a detailed study is underway and will be published in a forthcoming journal paper," which indicates that the present results are preliminary. The authors should either present the full experimental protocol or explicitly frame the current manuscript as a preliminary report.
minor comments (5)
  1. [Section IV-A] The dataset image size is given as "482 pixels"; the FER2013 images are 48×48 pixels. This appears to be a typographical error.
  2. [Section IV-B] The text mentions resizing images to "2242"; this is likely intended as 224×224 pixels.
  3. [Section III-B] There is a typo: "acheive" should be "achieve".
  4. [Figure 4 caption] The caption reads "Architecture designes for ConvNeXt and EmoNeXt"; "designes" should be "designs".
  5. [Section IV-C] The ordering of rows in Table II is partly non-monotonic. While this may reflect grouping by accuracy, the mixing makes visual scanning harder; a strictly sorted table with confidence intervals would be clearer.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an empirical architecture study whose accuracy claims are measurements, not predictions derived from fitted parameters or self-citation chains.

full rationale

The paper's central claim is that EmoNeXt-XLarge reaches 76.12% accuracy on FER2013, surpassing published state-of-the-art numbers. This is a measured benchmark result, not the output of a derivation that reduces to its own inputs. The architecture combines existing components (STN, SE blocks, ConvNeXt backbone) and a self-attention regularization term defined in Eq. (2) and combined in Eq. (3). No parameter is fitted to a subset of data and then reported as a prediction of a closely related quantity; the accuracy numbers are direct test-set measurements. The paper does not rely on self-citations for its load-bearing claims; the cited works are external and used for standard components or prior benchmark numbers. The main weaknesses of the paper are comparability and statistical significance: the claim of beating Segmentation VGG-19 by 0.15 percentage points is drawn from a mixed-protocol table, and the value of lambda in Eq. (3) is not reported. These are correctness/reproducibility concerns, not circularity. There is no step where an output is equivalent to an input by definition, no uniqueness theorem imported from the authors, and no ansatz smuggled in via self-citation. Therefore the circularity score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim depends on one unspecified loss weight lambda and several unstated training hyperparameters. It also rests on the reliability of the FER2013 labels, on beneficial transfer from ImageNet-22k pretraining, on the comparability of published baselines, and on the usefulness of the new variance penalty, none of which are independently demonstrated in this paper.

free parameters (3)
  • lambda (SA regularization weight) = not reported
    In Eq. (3), L_final = L_CE + lambda * L_SA. The value is never given, yet it sets the strength of the new regularizer and affects the reported accuracy.
  • batch size = not reported
    Used in all training runs on a T4 GPU but never stated; relevant to reproducing the optimization trajectory.
  • number of epochs = not reported
    Training length with cosine decay is never given; without it the reported accuracies cannot be reproduced exactly.
assumptions (4)
  • domain assumption FER2013 labels are reliable ground truth for the seven emotion classes.
    Invoked throughout Section IV-A; the entire evaluation treats the dataset labels as correct despite known label noise in FER2013.
  • domain assumption Pretrained ImageNet-22k weights transfer beneficially to FER2013.
    Section IV-B says pretrained ConvNeXt weights are incorporated; this transfer is assumed to help and is not tested against training from scratch.
  • domain assumption Baseline accuracies cited from other papers are accurate and directly comparable.
    Section IV-C and Table II compile numbers from different papers with different training setups; comparability is assumed.
  • ad hoc to paper The self-attention variance penalty improves generalization.
    Eq. (2) and Eq. (3) introduce this regularizer, but no ablation or theory establishes that it helps; it is a design choice specific to this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of EmoNeXt: an Adapted ConvNeXt for Facial Emotion Recognition." pith.science (2026). https://pith.science/paper/O6KW2DTP

@misc{pith2026250108199,
  author       = {Pith},
  title        = {Pith review of: EmoNeXt: an Adapted ConvNeXt for Facial Emotion Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O6KW2DTP}},
  note         = {Machine review of arXiv:2501.08199}
}
read the original abstract

Facial expressions play a crucial role in human communication serving as a powerful and impactful means to express a wide range of emotions. With advancements in artificial intelligence and computer vision, deep neural networks have emerged as effective tools for facial emotion recognition. In this paper, we propose EmoNeXt, a novel deep learning framework for facial expression recognition based on an adapted ConvNeXt architecture network. We integrate a Spatial Transformer Network (STN) to focus on feature-rich regions of the face and Squeeze-and-Excitation blocks to capture channel-wise dependencies. Moreover, we introduce a self-attention regularization term, encouraging the model to generate compact feature vectors. We demonstrate the superiority of our model over existing state-of-the-art deep learning models on the FER2013 dataset regarding emotion classification accuracy.

Figures

Figures reproduced from arXiv: 2501.08199 by the authors.

Figure 1
Figure 1. The architecture of a spatial transformer module. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 3
Figure 3. The architecture of the Squeeze-and-Excitation block. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 2
Figure 2. The ConvNeXt block. The authors of ConvNeXt have developed multiple versions, distinguished by variations in the number of channels (C) and blocks (B) within each stage. Here are the configurations: Tiny C = (96, 192, 384, 768) B = (3, 3, 9, 3) Small C = (96, 192, 384, 768) B = (3, 3, 27, 3) Base C = (128, 256, 512, 1024) B = (3, 3, 27, 3) Large C = (192, 384, 768, 1536) B = (3, 3, 27, 3) XLarge C = (256, 512, 1024,… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Architecture designes for ConvNeXt and EmoNeXt. [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Sample images from the FER2013 dataset. Additionally, we incorporate data augmentation techniques such as RandomCropping and RandomRotation to augment the training data, boosting the model’s ability to generalize. To prevent overfitting, we implement regularization sch…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 21 canonical work pages

  1. [14]

    Local learning with deep and handcrafted features for facial expression recognition,

    M.-I. Georgescu, R. T. Ionescu, and M. Popescu, “Local learning with deep and handcrafted features for facial expression recognition,” IEEE Access, vol. 7, pp. 64 827–64 836, 2019

    work page 2019

  2. [15]

    Local multi-head channel self- attention for facial expression recognition,

    R. Pecoraro, V . Basile, and V . Bono, “Local multi-head channel self- attention for facial expression recognition,” Information, vol. 13, no. 9, p. 419, 2022

    work page 2022

  3. [16]

    Ad-corre: Adaptive correlation-based loss for facial expression recognition in the wild,

    A. P. Fard and M. H. Mahoor, “Ad-corre: Adaptive correlation-based loss for facial expression recognition in the wild,” IEEE Access, vol. 10, pp. 26 756–26 768, 2022

    work page 2022

  4. [19]

    A novel facial emotion recognition model using segmentation vgg-19 architecture,

    S. Vignesh, M. Savithadevi, M. Sridevi, and R. Sridhar, “A novel facial emotion recognition model using segmentation vgg-19 architecture,” International Journal of Information Technology , pp. 1–11, 2023

    work page 2023

  5. [36]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein et al., “Imagenet large scale visual recognition challenge,” International journal of computer vision, vol. 115, pp. 211–252, 2015

    2015

  6. [40]

    Facial Expression Recognition with Deep Learning

    A. Khanzada, C. Bai, and F. T. Celepcikay, “Facial expression recogni- tion with deep learning,” arXiv preprint arXiv:2004.11823 , 2020

    work page Pith review arXiv 2004

  7. [1]

    Some referents and measures of nonverbal behavior,

    A. Mehrabian, “Some referents and measures of nonverbal behavior,” Behavior Research Methods & Instrumentation , vol. 1, no. 6, pp. 203– 207, 1968

    work page 1968

  8. [2]

    Ai and robotics to help older adults: Revisiting projects in search of lessons learned,

    G. Cortellessa, R. De Benedictis, F. Fracasso, A. Orlandini, A. Umbrico, and A. Cesta, “Ai and robotics to help older adults: Revisiting projects in search of lessons learned,” Paladyn, Journal of Behavioral Robotics , vol. 12, no. 1, pp. 356–378, 2021

    work page 2021

Show all 41 references
  1. [3]

    Raspberry pi assisted facial expression recognition framework for smart security in law-enforcement services,

    M. Sajjad, M. Nasir, F. U. M. Ullah, K. Muhammad, A. K. Sangaiah, and S. W. Baik, “Raspberry pi assisted facial expression recognition framework for smart security in law-enforcement services,” Information Sciences, vol. 479, pp. 416–431, 2019

    2019

  2. [4]

    Driver’s facial expression recognition in real- time for safe driving,

    M. Jeong and B. C. Ko, “Driver’s facial expression recognition in real- time for safe driving,” Sensors, vol. 18, no. 12, p. 4270, 2018

    2018

  3. [5]

    Recognition of facial expressions based on salient geometric features and support vector machines,

    D. Ghimire, J. Lee, Z.-N. Li, and S. Jeong, “Recognition of facial expressions based on salient geometric features and support vector machines,” Multimedia Tools and Applications, vol. 76, pp. 7921–7946, 2017

    2017

  4. [6]

    Facial expression recognition in image sequences using geometric deformation features and support vector machines,

    I. Kotsia and I. Pitas, “Facial expression recognition in image sequences using geometric deformation features and support vector machines,” IEEE transactions on image processing , vol. 16, no. 1, pp. 172–187, 2006

    2006

  5. [7]

    Facial expression recognition based on local binary patterns: A comprehensive study,

    C. Shan, S. Gong, and P. W. McOwan, “Facial expression recognition based on local binary patterns: A comprehensive study,” Image and vision Computing, vol. 27, no. 6, pp. 803–816, 2009

    2009

  6. [8]

    Facial expression recognition based on facial components detection and hog features,

    J. Chen, Z. Chen, Z. Chi, H. Fu et al. , “Facial expression recognition based on facial components detection and hog features,” in International workshops on electrical and computer engineering subfields , 2014, pp. 884–888

    2014

  7. [9]

    A convnet for the 2020s,

    Z. Liu, H. Mao, C.-Y . Wu, C. Feichtenhofer, T. Darrell, and S. Xie, “A convnet for the 2020s,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 11 976–11 986

    2022

  8. [10]

    Spatial transformer networks,

    M. Jaderberg, K. Simonyan, A. Zisserman et al. , “Spatial transformer networks,” Advances in neural information processing systems , vol. 28, 2015

    2015

  9. [11]

    Squeeze-and-excitation networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

    2018

  10. [12]

    Generalized cross entropy loss for training deep neural networks with noisy labels,

    Z. Zhang and M. Sabuncu, “Generalized cross entropy loss for training deep neural networks with noisy labels,” Advances in neural information processing systems, vol. 31, 2018

    2018

  11. [13]

    Chal- lenges in representation learning: A report on three machine learning contests,

    I. J. Goodfellow, D. Erhan, P. L. Carrier, A. Courville, M. Mirza, B. Hamner, W. Cukierski, Y . Tang, D. Thaler, D.-H. Lee et al., “Chal- lenges in representation learning: A report on three machine learning contests,” in Neural Information Processing: 20th International Confe...

    2013

  12. [17]

    Xception: Deep learning with depthwise separable convolu- tions,

    F. Chollet, “Xception: Deep learning with depthwise separable convolu- tions,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 1251–1258

    2017

  13. [18]

    Deep residual learning for image recognition,

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

    2016

  14. [20]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III

    2015

  15. [21]

    Springer, 2015, pp. 234–241

    2015

  16. [22]

    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

  17. [23]

    A deep-learning-based facial expression recognition method using textural features,

    M. Mukhopadhyay, A. Dey, and S. Kahali, “A deep-learning-based facial expression recognition method using textural features,” Neural Computing and Applications , vol. 35, no. 9, pp. 6499–6514, 2023

    2023

  18. [24]

    Role of zoning in facial expression using deep learning,

    T. Shahzad, K. Iqbal, M. A. Khan, N. Iqbal et al. , “Role of zoning in facial expression using deep learning,” IEEE Access , vol. 11, pp. 16 493–16 508, 2023

    2023

  19. [25]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

    2010 arXiv

  20. [26]

    Rectified linear units improve restricted boltz- mann machines,

    V . Nair and G. E. Hinton, “Rectified linear units improve restricted boltz- mann machines,” in Proceedings of the 27th international conference on machine learning (ICML-10) , 2010, pp. 807–814

    2010

  21. [27]

    Gaussian error linear units (gelus),

    D. Hendrycks and K. Gimpel, “Gaussian error linear units (gelus),” arXiv preprint arXiv:1606.08415, 2016

    2016 arXiv

  22. [28]

    Batch renormalization: Towards reducing minibatch depen- dence in batch-normalized models,

    S. Ioffe, “Batch renormalization: Towards reducing minibatch depen- dence in batch-normalized models,” Advances in neural information processing systems, vol. 30, 2017

    2017

  23. [29]

    Layer normalization,

    J. L. Ba, J. R. Kiros, and G. E. Hinton, “Layer normalization,” arXiv preprint arXiv:1607.06450, 2016

    2016 arXiv

  24. [30]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017

    2017

  25. [31]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101 , 2017

    2017 arXiv

  26. [32]

    Deep networks with stochastic depth,

    G. Huang, Y . Sun, Z. Liu, D. Sedra, and K. Q. Weinberger, “Deep networks with stochastic depth,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14 . Springer, 2016, pp. 646–661

    2016

  27. [33]

    Rethinking the inception architecture for computer vision,

    C. Szegedy, V . Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, “Rethinking the inception architecture for computer vision,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2016, pp. 2818–2826

    2016

  28. [34]

    Acceleration of stochastic approxima- tion by averaging,

    B. T. Polyak and A. B. Juditsky, “Acceleration of stochastic approxima- tion by averaging,” SIAM journal on control and optimization , vol. 30, no. 4, pp. 838–855, 1992

    1992

  29. [35]

    Mixed precision training,

    P. Micikevicius, S. Narang, J. Alben, G. Diamos, E. Elsen, D. Garcia, B. Ginsburg, M. Houston, O. Kuchaiev, G. Venkatesh et al. , “Mixed precision training,” arXiv preprint arXiv:1710.03740 , 2017

    2017 arXiv

  30. [37]

    Deep learning approaches for facial emotion recognition: A case study on fer-2013,

    P. Giannopoulos, I. Perikos, and I. Hatzilygeroudis, “Deep learning approaches for facial emotion recognition: A case study on fer-2013,” Advances in Hybridization of Intelligent Methods: Models, Systems and Applications, pp. 1–16, 2018

    2013

  31. [38]

    Deep-emotion: Facial ex- pression recognition using attentional convolutional network,

    S. Minaee, M. Minaei, and A. Abdolrashidi, “Deep-emotion: Facial ex- pression recognition using attentional convolutional network,” Sensors, vol. 21, no. 9, p. 3046, 2021

    2021

  32. [39]

    Facial emotion recognition: State of the art performance on fer2013,

    Y . Khaireddin and Z. Chen, “Facial emotion recognition: State of the art performance on fer2013,” arXiv preprint arXiv:2105.03588 , 2021

    2021 arXiv

  33. [41]

    Facial expression recognition using residual masking network,

    L. Pham, T. H. Vu, and T. A. Tran, “Facial expression recognition using residual masking network,” in 2020 25Th international conference on pattern recognition (ICPR) . IEEE, 2021, pp. 4513–4519

    2020

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.