Pith. sign in

REVIEW 3 major objections 4 minor 4 cited by

CoEmoGen: Towards Semantically-Coherent and Scalable Emotional Image Content Generation

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read CoEmoGen claims that emotion-guided image generation works best when driven by rich, emotion-triggering captions and a hierarchical low-rank adaptation that separates shared and emotion-specific features.

desk verdict A plausible new pipeline for emotion-guided image generation, but the supplied full text is mojibake, so the central empirical claims cannot be verified. read the letter →

arxiv 2508.03535 v1 pith:UJADJALU submitted 2025-08-05 cs.CV

Kaishen Yuan , Yuting Zhang , Shang Gao , Yijie Zhu , Wenshuo Chen , Yutao Yue This is my paper
classification cs.CV
keywords emotionalimagecontentgenerationtext-to-imagediffusionhierarchicallow-rankadaptationmultimodallargelanguagemodelssemanticcoherencefaithfulnessEmoArtdataset
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

The paper tries to establish that Emotional Image Content Generation fails when models depend on word-level attribute labels, which are ambiguous and do not scale. It proposes CoEmoGen, a pipeline that constructs emotion-triggering captions with multimodal large language models and models emotion with a Hierarchical Low-Rank Adaptation (HiLoRA) module. If the claim holds, CoEmoGen produces images that are both emotionally faithful and semantically coherent, outperforming prior EICG methods and general text-to-image models. The paper also introduces EmoArt, a large-scale dataset of emotionally evocative artistic images, to support scalable emotion-driven creation.

What carries the argument

The central mechanism is HiLoRA, a hierarchical low-rank adaptation module that injects emotion knowledge into a frozen diffusion model at two levels: a shared low-rank branch captures polarity-level features common to emotions of the same valence, while emotion-specific branches capture high-level semantics unique to each emotion category. This is paired with captions produced by a multimodal large language model that describes emotion-triggering content in context, replacing word-level attribute labels. Together they turn a generic text-to-image diffusion model into an emotion-controllable generator.

What would settle it

Compare HiLoRA against a standard LoRA with matched parameter count on the same emotion-to-image benchmark; if per-emotion emotional faithfulness and semantic coherence scores are statistically indistinguishable for same-polarity emotions, the hierarchy is not contributing. Inspect the shared low-rank branch by ablating it: if outputs across all positive or all negative emotions are unchanged, the polarity-shared factorization fails.

Watch

Extended reading notes

Core claim

CoEmoGen's central claim is that emotional image generation should be driven by context-rich, emotion-triggering text captions rather than word-level attribute labels, and that the emotion itself can be factorized into polarity-shared low-level features and emotion-specific high-level semantics. With these two components, a text-to-image diffusion model can generate images that reliably express a specified emotion category while staying semantically coherent. The paper reports that CoEmoGen demonstrates superiority in emotional faithfulness and semantic coherence over prior EICG methods and general text-to-image models, as evidenced by quantitative metrics, qualitative examples, and user studies.

Load-bearing premise

The load-bearing premise is that emotions can be decomposed into polarity-shared low-level features and emotion-specific high-level semantics, and that HiLoRA can learn that split; if the decomposition is not real or not separable in the model, the module has no principled advantage over standard LoRA.

Editorial extensions

If this is right

  • If CoEmoGen is correct, emotion-driven creative tools can generate contextually rich, emotionally faithful images without hand-written attribute labels, improving scalability.
  • MLLM-based captioning can serve as a general front-end for abstract-concept generation beyond emotions, such as mood, atmosphere, or narrative tone.
  • The HiLoRA factorization suggests that polarity-shared features can be reused across emotion categories, potentially reducing parameters or enabling efficient adaptation to new emotions.
  • The EmoArt dataset can serve as a benchmark for emotional image generation and as a resource for artistic creation and analysis.

Reading between the lines

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

  • The paper does not demonstrate that the polarity-shared decomposition is the only valid factorization; a natural extension would test whether HiLoRA's shared low-rank branch transfers across emotion categories in a zero-shot setting.
  • The same caption-plus-hierarchical-LoRA recipe could be applied to other abstract attributes such as aesthetic style or narrative emotion, where word-level labels are equally insufficient.
  • The reported user-study gains could be strengthened by connecting generated images to physiological or behavioral measures of emotion, rather than relying solely on self-reported recognition.
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

3 major / 4 minor

Summary. The paper proposes CoEmoGen, a pipeline for emotional image content generation that combines MLLM-generated emotion-focused captions with a Hierarchical Low-Rank Adaptation (HiLoRA) module, and introduces EmoArt, a large-scale dataset of emotionally evocative artistic images. The abstract claims that CoEmoGen achieves superiority over prior EICG methods and general text-to-image models in emotional faithfulness and semantic coherence, supported by quantitative, qualitative, and user-study evidence. However, the submitted full text is entirely corrupted mojibake, so the method description, equations, tables, figures, experiments, and user-study details are all unreadable. The review is therefore based on the abstract alone and on the general plausibility of the proposed approach.

Significance. If the claims are substantiated, CoEmoGen would provide a noteworthy contribution to emotional image generation, and EmoArt could serve as a valuable benchmark and inspiration source for the community. The combination of MLLM-guided captioning with a psychologically motivated LoRA variant is intriguing and potentially useful. The authors also commit to releasing code and data, which is commendable and would aid reproducibility. However, the current submission artifact prevents any substantive technical assessment; the significance of the work remains conditional on a legible, verifiable version of the manuscript.

major comments (3)
  1. [Full text (all sections)] The submitted full text is entirely corrupted mojibake; not a single section, equation, table, figure caption, baseline list, metric, or experimental number is legible. The abstract's central claim that 'extensive experiments demonstrate CoEmoGen's superiority' cannot be checked in any way from the provided artifact. This is a load-bearing defect, not a stylistic issue, because the manuscript's contributions are empirical and the evaluation is promised but unavailable. Please provide a readable PDF version so that the experimental evidence can be reviewed.
  2. [Abstract, HiLoRA motivation] The HiLoRA module is motivated by a psychological decomposition of emotions into polarity-shared low-level features and emotion-specific high-level semantics, but the abstract gives no citation or empirical support for this decomposition. The current text asserts the split as an inspiration rather than demonstrating that it holds, and the corrupted full text prevents checking whether the paper provides evidence elsewhere. If this assumption is load-bearing for the proposed module, the resubmission should include a concrete justification or ablation that shows HiLoRA outperforms standard LoRA under controlled conditions.
  3. [Abstract, experimental evidence] The abstract mentions only 'quantitative, qualitative, and user study perspectives' without listing any baselines, datasets, metrics, error bars, or significance tests. Because the full text is unreadable, the claimed 'superiority' over prior methods is unfalsifiable as presented. The resubmission must include legible experimental sections with named comparison methods, evaluated emotion categories, evaluation metrics, and proper statistical reporting for the user study.
minor comments (4)
  1. [Abstract] The abstract does not state which emotion categories are handled or which datasets are used for evaluation; please add this information so that readers can assess the scope of the claimed generalization.
  2. [Abstract] The phrase 'endless inspiration' is informal for a technical paper; consider rewording to something like 'a large-scale resource to support emotion-driven artistic creation'.
  3. [Abstract] The paper promises code and dataset release, but the abstract does not specify a license for the EmoArt dataset; please include this detail in the resubmission.
  4. [Method description (full text)] The abstract refers to multimodal large language models without specifying which MLLM is used; please state the specific model and, if the same MLLM is used for captioning and for evaluation, address the potential for evaluation bias.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity is identifiable from the readable abstract; the full text is corrupted, so the empirical claims cannot be verified, but nothing in the available text shows a derivation reducing to its own inputs.

full rationale

The only fully readable portion is the abstract. It describes a pipeline: MLLM-generated emotion-focused captions provide semantic guidance, and a HiLoRA module models polarity-shared low-level features and emotion-specific high-level semantics. There are no equations, no explicit fitting procedures, and no visible self-citations in the supplied text, so none of the enumerated circularity patterns (self-definitional fitting, fitted-input-as-prediction, load-bearing self-citation, imported uniqueness, ansatz smuggled via citation, or renaming a known result) can be established. The claim of superiority rests on unreported experiments, but the inability to inspect those experiments is a reviewability and correctness-risk concern, not evidence of circularity. Under the hard rule that circularity must be demonstrated by quotation and specific reduction, and the default expectation that most papers are not circular, the appropriate finding is no significant circularity with score 0.

Assumptions & free parameters 1 free parameters · 2 assumptions · 0 invented entities

From the abstract alone, the model relies on two substantive domain assumptions: the psychological decomposition of emotions and the superiority of MLLM-generated captions over word labels. All other components (diffusion models, MLLMs, LoRA) are standard tools. The rank/alpha of HiLoRA are likely tuned hyperparameters, but their values are not disclosed in the abstract.

free parameters (1)
  • HiLoRA rank and alpha hyperparameters
    Not stated in abstract; typical LoRA modules require choosing rank and scaling, and these would be tuned on validation data, affecting the emotion-specific vs shared tradeoff.
assumptions (2)
  • domain assumption Emotions can be decomposed into polarity-shared low-level features and emotion-specific high-level semantics.
    Stated in abstract as the psychological insight underlying HiLoRA design. If false, the module has no principled basis.
  • domain assumption MLLM-generated emotion-triggering captions are more effective than word-level attribute labels for guiding diffusion models.
    Core motivation of the paper. The abstract asserts context-rich captions improve semantic coherence, but provides no evidence within the abstract.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CoEmoGen: Towards Semantically-Coherent and Scalable Emotional Image Content Generation." pith.science (2026). https://pith.science/paper/UJADJALU

@misc{pith2026250803535,
  author       = {Pith},
  title        = {Pith review of: CoEmoGen: Towards Semantically-Coherent and Scalable Emotional Image Content Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UJADJALU}},
  note         = {Machine review of arXiv:2508.03535}
}
read the original abstract

Emotional Image Content Generation (EICG) aims to generate semantically clear and emotionally faithful images based on given emotion categories, with broad application prospects. While recent text-to-image diffusion models excel at generating concrete concepts, they struggle with the complexity of abstract emotions. There have also emerged methods specifically designed for EICG, but they excessively rely on word-level attribute labels for guidance, which suffer from semantic incoherence, ambiguity, and limited scalability. To address these challenges, we propose CoEmoGen, a novel pipeline notable for its semantic coherence and high scalability. Specifically, leveraging multimodal large language models (MLLMs), we construct high-quality captions focused on emotion-triggering content for context-rich semantic guidance. Furthermore, inspired by psychological insights, we design a Hierarchical Low-Rank Adaptation (HiLoRA) module to cohesively model both polarity-shared low-level features and emotion-specific high-level semantics. Extensive experiments demonstrate CoEmoGen's superiority in emotional faithfulness and semantic coherence from quantitative, qualitative, and user study perspectives. To intuitively showcase scalability, we curate EmoArt, a large-scale dataset of emotionally evocative artistic images, providing endless inspiration for emotion-driven artistic creation. The dataset and code are available at https://github.com/yuankaishen2001/CoEmoGen.

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Oracle Noise: Faster Semantic Spherical Alignment for Interpretable Latent Optimization

    cs.CV 2026-04 unverdicted novelty 7.0 of 10

    Oracle Noise optimizes diffusion model noise on a Riemannian hypersphere guided by key prompt words to preserve the Gaussian prior, eliminate norm inflation, and achieve faster semantic alignment than Euclidean methods.

  2. $Z^2$-Sampling: Zero-Cost Zigzag Trajectories for Semantic Alignment in Diffusion Models

    cs.CV 2026-04 unverdicted novelty 7.0 of 10

    Z²-Sampling implicitly realizes zero-cost zigzag trajectories for curvature-aware semantic alignment in diffusion models by reducing multi-step paths via operator dualities and temporal caching while synthesizing a di...

  3. What Can I Edit? Open-Ended Strategy Discovery and the Emotion Editability Landscape

    cs.CV 2026-07 conditional novelty 6.0 of 10

    An open-ended 'what can I edit' planner that discovers image-specific emotional affordances before editing produces edits humans strongly prefer over fixed-taxonomy emotional image editors, while standard emotion clas...

  4. EmoStyle: Affective Conditioning of Style-Specialist Experts for Emotional Image Generation

    cs.CV 2026-07 conditional novelty 5.0 of 10

    EmoStyle injects LLM-inferred valence-arousal and emotion labels into Z-Image via AdaLN-style residual modulation over style-bucket LoRA experts, plus VLM candidate ranking, and ranked first on AffectiveArt Track 1.

Reference graph

Works this paper leans on

60 extracted references · 33 canonical work pages · cited by 4 Pith papers

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Bai, Z.; Wang, P.; Xiao, T.; He, T.; Han, Z.; Zhang, Z.; and Shou, M. Z. 2024. Hallucination of multimodal large language models: A survey. arXiv preprint arXiv:2404.18930

    arXiv 2024

  4. [4]

    Betker, J.; Goh, G.; Jing, L.; Brooks, T.; Wang, J.; Li, L.; Ouyang, L.; Zhuang, J.; Lee, J.; Guo, Y.; et al. 2023. Improving image generation with better captions. Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf, 2(3): 8

    2023

  5. [5]

    Borth, D.; Ji, R.; Chen, T.; Breuel, T.; and Chang, S.-F. 2013. Large-scale visual sentiment ontology and detectors using adjective noun pairs. In Proceedings of the 21st ACM international conference on Multimedia, 223--232

    work page 2013

  6. [6]

    Brosch, T.; Pourtois, G.; and Sander, D. 2010. The perception and categorisation of emotional stimuli: A review. Cognition and emotion, 24(3): 377--400

    work page 2010

  7. [7]

    Chen, T.; Xiong, W.; Zheng, H.; and Luo, J. 2020. Image sentiment transfer. In Proceedings of the 28th ACM International Conference on Multimedia, 4407--4415

    work page 2020

  8. [8]

    Chen, W.; Xiao, H.; Zhang, E.; Hu, L.; Wang, L.; Liu, M.; and Chen, C. 2024. Sato: Stable text-to-motion framework. In Proceedings of the 32nd ACM International Conference on Multimedia, 6989--6997

    work page 2024

Show all 60 references
  1. [9]

    Chen, W.; Yu, K.; Jia, H.; Yuan, K.; Tian, B.; Lai, S.; Xiao, H.; Zhang, E.; Wang, L.; and Yue, Y. 2025. ANT: Adaptive Neural Temporal-Aware Text-to-Motion Model. arXiv preprint arXiv:2506.02452

    2025 arXiv

  2. [10]

    Dhariwal, P.; and Nichol, A. 2021. Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 34: 8780--8794

    2021

  3. [11]

    H.; Chechik, G.; and Cohen-Or, D

    Gal, R.; Alaluf, Y.; Atzmon, Y.; Patashnik, O.; Bermano, A. H.; Chechik, G.; and Cohen-Or, D. 2022. An image is worth one word: Personalizing text-to-image generation using textual inversion. arXiv preprint arXiv:2208.01618

    2022 arXiv

  4. [12]

    Heusel, M.; Ramsauer, H.; Unterthiner, T.; Nessler, B.; and Hochreiter, S. 2017. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30

    2017

  5. [13]

    Ho, J.; Jain, A.; and Abbeel, P. 2020. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 6840--6851

    2020

  6. [14]

    Ho, J.; and Salimans, T. 2022. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598

    2022 arXiv

  7. [15]

    J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W.; et al

    Hu, E. J.; Shen, Y.; Wallis, P.; Allen-Zhu, Z.; Li, Y.; Wang, S.; Wang, L.; Chen, W.; et al. 2022. Lora: Low-rank adaptation of large language models. ICLR, 1(2): 3

    2022

  8. [16]

    Hu, Z.; Yuan, K.; Liu, X.; Yu, Z.; Zong, Y.; Shi, J.; Yue, H.; and Yang, J. 2025. Feallm: Advancing facial emotion analysis in multimodal large language models with emotional synergy and reasoning. arXiv preprint arXiv:2505.13419

    2025 arXiv

  9. [17]

    M.; Bader, J.; Alaniz, S.; Schmid, C.; and Akata, Z

    Kim, J. M.; Bader, J.; Alaniz, S.; Schmid, C.; and Akata, Z. 2024. Datadream: Few-shot guided dataset generation. In European Conference on Computer Vision, 252--268. Springer

    2024

  10. [18]

    Kumari, N.; Zhang, B.; Zhang, R.; Shechtman, E.; and Zhu, J.-Y. 2023. Multi-concept customization of text-to-image diffusion. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 1931--1941

    2023

  11. [19]

    Kuznetsova, A.; Rom, H.; Alldrin, N.; Uijlings, J.; Krasin, I.; Pont-Tuset, J.; Kamali, S.; Popov, S.; Malloci, M.; Kolesnikov, A.; et al. 2020. The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale. Internationa...

    2020

  12. [20]

    Li, Y.; Lai, Z.; Bao, W.; Tan, Z.; Dao, A.; Sui, K.; Shen, J.; Liu, D.; Liu, H.; and Kong, Y. 2025. Visual Large Language Models for Generalized and Specialized Applications. arXiv preprint arXiv:2501.02765

    2025 arXiv

  13. [21]

    Liu, D.; Jiang, Y.; Pei, M.; and Liu, S. 2018. Emotional image color transfer via deep learning. Pattern Recognition Letters, 110: 16--22

    2018

  14. [22]

    Liu, X.; Yuan, K.; Niu, X.; Shi, J.; Yu, Z.; Yue, H.; and Yang, J. 2024 a . Multi-scale promoted self-adjusting correlation learning for facial action unit detection. IEEE Transactions on Affective Computing

    2024

  15. [23]

    Liu, X.; Zhang, Y.; Yu, Z.; Lu, H.; Yue, H.; and Yang, J. 2024 b . rppg-mae: Self-supervised pretraining with masked autoencoders for remote physiological measurements. IEEE Transactions on Multimedia, 26: 7278--7293

    2024

  16. [24]

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

    2017 arXiv

  17. [25]

    Lu, H.; Niu, X.; Wang, J.; Wang, Y.; Hu, Q.; Tang, J.; Zhang, Y.; Yuan, K.; Huang, B.; Yu, Z.; et al. 2024. Gpt as psychologist? preliminary evaluations for gpt-4v on visual affective computing. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogniti...

    2024

  18. [26]

    Machajdik, J.; and Hanbury, A. 2010. Affective image classification using features inspired by psychology and art theory. In Proceedings of the 18th ACM international conference on Multimedia, 83--92

    2010

  19. [27]

    A.; Fredrickson, B

    Mikels, J. A.; Fredrickson, B. L.; Larkin, G. R.; Lindberg, C. M.; Maglio, S. J.; and Reuter-Lorenz, P. A. 2005. Emotional category data on images from the International Affective Picture System. Behavior research methods, 37: 626--630

    2005

  20. [28]

    Minsky, M. 2007. The emotion machine: Commonsense thinking, artificial intelligence, and the future of the human mind. Simon and Schuster

    2007

  21. [29]

    Nichol, A.; Dhariwal, P.; Ramesh, A.; Shyam, P.; Mishkin, P.; McGrew, B.; Sutskever, I.; and Chen, M. 2021. Glide: Towards photorealistic image generation and editing with text-guided diffusion models. arXiv preprint arXiv:2112.10741

    2021 arXiv

  22. [30]

    Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury, J.; Chanan, G.; Killeen, T.; Lin, Z.; Gimelshein, N.; Antiga, L.; et al. 2019. Pytorch: An imperative style, high-performance deep learning library. Advances in neural information processing systems, 32

    2019

  23. [31]

    T.; Zaidel, D

    Pearce, M. T.; Zaidel, D. W.; Vartanian, O.; Skov, M.; Leder, H.; Chatterjee, A.; and Nadal, M. 2016. Neuroaesthetics: The cognitive neuroscience of aesthetic experience. Perspectives on psychological science, 11(2): 265--279

    2016

  24. [32]

    Peebles, W.; and Xie, S. 2023. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF international conference on computer vision, 4195--4205

    2023

  25. [33]

    Peng, K.-C.; Chen, T.; Sadovnik, A.; and Gallagher, A. C. 2015. A mixed bag of emotions: Model, predict, and transfer emotion distributions. In Proceedings of the IEEE conference on computer vision and pattern recognition, 860--868

    2015

  26. [34]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PmLR

    2021

  27. [35]

    Rao, T.; Li, X.; and Xu, M. 2020. Learning multi-level deep representations for image emotion classification. Neural processing letters, 51: 2043--2061

    2020

  28. [36]

    Rombach, R.; Blattmann, A.; Lorenz, D.; Esser, P.; and Ommer, B. 2022. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 10684--10695

    2022

  29. [37]

    Ruiz, N.; Li, Y.; Jampani, V.; Pritch, Y.; Rubinstein, M.; and Aberman, K. 2023. Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 22500--22510

    2023

  30. [38]

    L.; Ghasemipour, K.; Gontijo Lopes, R.; Karagol Ayan, B.; Salimans, T.; et al

    Saharia, C.; Chan, W.; Saxena, S.; Li, L.; Whang, J.; Denton, E. L.; Ghasemipour, K.; Gontijo Lopes, R.; Karagol Ayan, B.; Salimans, T.; et al. 2022. Photorealistic text-to-image diffusion models with deep language understanding. Advances in neural information processing syste...

    2022

  31. [39]

    Song, J.; Meng, C.; and Ermon, S. 2020. Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502

    2020 arXiv

  32. [40]

    Sun, S.; Jia, J.; Wu, H.; Ye, Z.; and Xing, J. 2023. Msnet: A deep architecture using multi-sentiment semantics for sentiment-aware image style transfer. In ICASSP 2023-2023 IEEE international conference on acoustics, speech and signal processing (ICASSP), 1--5. IEEE

    2023

  33. [41]

    Tao, J.; and Tan, T. 2005. Affective computing: A review. In International Conference on Affective computing and intelligent interaction, 981--995. Springer

    2005

  34. [42]

    Tian, K.; Jiang, Y.; Yuan, Z.; Peng, B.; and Wang, L. 2025. Visual autoregressive modeling: Scalable image generation via next-scale prediction. Advances in neural information processing systems, 37: 84839--84865

    2025

  35. [43]

    Wang, X.; Jia, J.; and Cai, L. 2013. Affective image adjustment with a single word. The Visual Computer, 29: 1121--1133

    2013

  36. [44]

    Weng, S.; Zhang, P.; Chang, Z.; Wang, X.; Li, S.; and Shi, B. 2023. Affective image filter: Reflecting emotions from text to images. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 10810--10819

    2023

  37. [45]

    J.; Klupp, E.; Weyers, P.; Pauli, P.; Weise, D.; Zeller, D.; Classen, J.; and M \"u hlberger, A

    Wieser, M. J.; Klupp, E.; Weyers, P.; Pauli, P.; Weise, D.; Zeller, D.; Classen, J.; and M \"u hlberger, A. 2012. Reduced early visual emotion discrimination as an index of diminished emotion processing in Parkinson’s disease?--Evidence from event-related brain potentials. Cor...

    2012

  38. [46]

    Xing, B.; Yu, Z.; Liu, X.; Yuan, K.; Ye, Q.; Xie, W.; Yue, H.; Yang, J.; and K \"a lvi \"a inen, H. 2024. Emo-llama: Enhancing facial emotion understanding with instruction tuning. arXiv preprint arXiv:2408.11424

    2024 arXiv

  39. [47]

    Xing, B.; Yuan, K.; Yu, Z.; Liu, X.; and K \"a lvi \"a inen, H. 2025. AU-TTT: Vision Test-Time Training model for Facial Action Unit Detection. arXiv preprint arXiv:2503.23450

    2025 arXiv

  40. [48]

    Yang, J.; Feng, J.; and Huang, H. 2024. EmoGen: Emotional image content generation with text-to-image diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 6358--6368

    2024

  41. [49]

    Yang, J.; Huang, Q.; Ding, T.; Lischinski, D.; Cohen-Or, D.; and Huang, H. 2023. Emoset: A large-scale visual emotion dataset with rich attributes. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 20383--20394

    2023

  42. [50]

    Yang, J.; Li, J.; Wang, X.; Ding, Y.; and Gao, X. 2021. Stimuli-aware visual emotion analysis. IEEE Transactions on Image Processing, 30: 7432--7445

    2021

  43. [51]

    L.; and Wang, L

    Yang, J.; She, D.; Sun, M.; Cheng, M.-M.; Rosin, P. L.; and Wang, L. 2018. Visual sentiment prediction based on automatic discovery of affective regions. IEEE Transactions on Multimedia, 20(9): 2513--2525

    2018

  44. [52]

    Yuan, K.; Yu, Z.; Liu, X.; Xie, W.; Yue, H.; and Yang, J. 2024. Auformer: Vision transformers are parameter-efficient facial action unit detectors. In European Conference on Computer Vision, 427--445. Springer

    2024

  45. [53]

    Zhang, L.; Rao, A.; and Agrawala, M. 2023. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF international conference on computer vision, 3836--3847

    2023

  46. [54]

    A.; Shechtman, E.; and Wang, O

    Zhang, R.; Isola, P.; Efros, A. A.; Shechtman, E.; and Wang, O. 2018. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition, 586--595

    2018

  47. [55]

    Zhang, Y.; Lu, H.; Hu, Q.; Wang, Y.; Yuan, K.; Liu, X.; and Wu, K. 2025 a . Period-LLM: Extending the Periodic Capability of Multimodal Large Language Model. In Proceedings of the Computer Vision and Pattern Recognition Conference, 29237--29247

    2025

  48. [56]

    Zhang, Y.; Lu, H.; Liu, X.; Chen, Y.; and Wu, K. 2025 b . Advancing generalizable remote physiological measurement through the integration of explicit and implicit prior knowledge. IEEE Transactions on Image Processing

    2025

  49. [57]

    Zhang, Y.; Yuan, K.; Lu, H.; Yue, Y.; Chen, J.; and Wu, K. 2025 c . MedTVT-R1: A Multimodal LLM Empowering Medical Reasoning and Diagnosis. arXiv preprint arXiv:2506.18512

    2025 arXiv

  50. [58]

    W.; and Keutzer, K

    Zhao, S.; Yao, X.; Yang, J.; Jia, G.; Ding, G.; Chua, T.-S.; Schuller, B. W.; and Keutzer, K. 2021. Affective image content analysis: Two decades review and new perspectives. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(10): 6729--6751

    2021

  51. [59]

    Zhou, B.; Lapedriza, A.; Khosla, A.; Oliva, A.; and Torralba, A. 2017. Places: A 10 million image database for scene recognition. IEEE transactions on pattern analysis and machine intelligence, 40(6): 1452--1464

    2017

  52. [60]

    Zhu, C.; Li, K.; Ma, Y.; He, C.; and Li, X. 2024. Multibooth: Towards generating all your concepts in an image from text. arXiv preprint arXiv:2404.14239

    2024 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.