Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Semi-Supervised Adaptation of Diffusion Models for Handwritten Text Generation

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A latent diffusion model can learn a new, unseen writer's style from unlabeled images via masked-autoencoder embeddings; semi-supervised training on the new dataset lowers generated-data character error rate from 9.99% to 6.61%.

desk verdict Useful extension of Wordstylist to unseen writers via MAE style embeddings, but the headline RIMES gain is suspect because the semi-supervised adaptation set and the CER-train evaluation set are the same RIMES train images. read the letter →

arxiv 2412.15853 v1 pith:NURFG7AE submitted 2024-12-20 cs.CV

Kai Brandenbusch This is my paper · ORCID
classification cs.CV
keywords handwrittentextgenerationlatentdiffusionmodelsmaskedautoencoderstyleembeddingsemi-supervisedlearningdomainadaptationclassifier-freeguidancehandwritingrecognition
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 aims to make latent diffusion models for handwritten text generation (HTG) usable for new, previously unseen handwriting datasets that come without word transcriptions. It claims that a masked autoencoder trained only to reconstruct randomly masked patches produces writer style embeddings when its patch vectors are averaged, so the diffusion model can imitate writers it never saw during training. Building on the Wordstylist architecture, the paper adds a content encoder that injects the style embedding into the text conditioning in several ways, uses classifier-free guidance, and proposes a semi-supervised training scheme that adds unlabeled target-domain images with masked text conditioning. The payoff would be practical: a handwriting recognition system could be adapted to a new corpus (here RIMES, with its own language and distribution) using only writer-grouped, unlabeled images, cutting the character error rate of synthetic training data from 9.99% to 6.61%.

What carries the argument

The load-bearing component is the masked autoencoder (MAE) style encoder, which computes a writer embedding by averaging the patch-level latent vectors of $K$ example images: $\mathrm{SE}(w) = \frac{1}{K N}\sum_{i=1}^{K}\sum_{j=1}^{N}\mathrm{SE}({}^i x_p^j)$. The paper assumes the unmasking reconstruction objective forces the encoder to store stroke and calligraphic information, not just letter content. Around this encoder, the content encoder offers six ways to inject the style vector into the text conditioning (TP, TPL, CP, TA, CA, TS), with TS—adding the projected style vector to the timestep embedding—performing best. The semi-supervised trick is to replace the text-conditioning sequence with the same mask tokens used for classifier-free guidance, so unlabeled target images contribute style information while the model keeps learning from labeled examples.

What would settle it

Train the masked autoencoder exactly as described, then measure writer-identification accuracy on the averaged embeddings for held-out writers: if a linear classifier trained on writer labels performs near chance, or if the embeddings cluster by word content rather than by writer, the central mechanism is not carrying style information and the unseen-writer and semi-supervised results would need another explanation.

Watch

Extended reading notes

Core claim

The paper's central claim is that a latent diffusion model conditioned on a masked-autoencoder style embedding can generate readable word images in the style of writers never seen during training, and that the same model can be adapted to an entirely new dataset without any transcriptions. In the RIMES experiment, a model trained only on IAM achieves a character error rate of 9.99% when asked to replicate RIMES training data, whereas the semi-supervised model—trained on labeled IAM plus unlabeled RIMES images whose text conditioning is replaced by mask tokens—reaches 6.61%, approaching the 3.25% of a model trained with full RIMES labels. The paper presents this as the first demonstration of semi-supervised training for DDPM-based HTG and as a step toward using generated images to adapt downstream recognition models to new domains.

Load-bearing premise

The method collapses if a masked autoencoder trained only to reconstruct masked patches does not actually encode calligraphic style in its averaged patch embeddings; the paper states this as an assumption and never directly measures whether the embeddings distinguish writers better than they distinguish content.

Editorial extensions

If this is right

  • An HTR system can be adapted to a new handwriting dataset, even in a different language, using only writer-grouped images and no transcriptions.
  • Semi-supervised training improves the synthetic training data for the new dataset, cutting CER-train from 9.99% to 6.61%, approaching the 3.25% achieved with full labels.
  • Classifier-free guidance with a moderate scale improves generation fidelity for known datasets, while for unseen datasets too-large guidance can hurt.
  • Unseen writers can be imitated from around ten example images, which is practical when one page implies one writer.
  • The choice of MAE training data (IAM-only vs. IAM+RIMES) has little influence on the final RIMES generation quality, suggesting the style representation transfers across datasets.

Reading between the lines

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

  • The paper never directly measures whether the MAE embeddings encode style; a writer-discrimination test on the averaged embeddings would validate or undermine the mechanism without retraining the diffusion model.
  • Part of the semi-supervised gain may come from matching the target dataset's image-level statistics (ink, background, preprocessing) rather than from true calligraphic transfer; if so, a simpler domain-adaptation module might achieve part of the improvement.
  • The semi-supervised scheme could be extended to cases where even the writer grouping is unknown, replacing the style embedding with a nearest-neighbor style estimate, which the paper leaves as future work.
  • Retraining the frozen VAE on handwritten documents, rather than using a natural-image VAE, could combine with this semi-supervised scheme and potentially close the remaining gap to fully supervised generation.
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 extends the Wordstylist latent diffusion model for handwritten text generation (HTG) by introducing a masked-autoencoder-based style encoder, several content-encoder variants for combining text and style conditioning, classifier-free guidance, and a semi-supervised training scheme for adapting the model to a new unlabeled dataset. The method is evaluated on IAM as the fully labeled source dataset and RIMES as the target dataset, using downstream HTR character error rates on generated training images as the main metric. The paper reports that semi-supervised adaptation reduces RIMES CER-train from about 9.99% (IAM-only) to 6.61% (semi-supervised), and concludes that the approach enables generation of usable training images for previously unseen datasets.

Significance. If the central claim holds, the work is significant for practical HTG applications: it would allow adapting a diffusion-based handwriting generator to a new corpus using only unlabeled writer-grouped images, with downstream HTR training as the payoff. The paper's strengths include a clear architectural description, a broad ablation of style-conditioning placements and guidance scales, and quantitative comparisons against two public reference implementations (Wordstylist and DiffusionPen), which is more extensive than many HTG reports. The main limitation is that the headline RIMES improvement is measured on the same image set used for adaptation, so the external validity of the 'unseen dataset' claim is not yet established. The MAE style-fidelity assumption is also not directly validated. These gaps are fixable with additional experiments, but they are load-bearing for the paper's main claim.

major comments (4)
  1. [Sec. 5.2/4.5, Table 9] The central RIMES improvement may reflect overlap between the adaptation set and the evaluation set. CER-train is defined in Sec. 5.2 as replicating the train set, i.e., generating the same text/style conditioning pairs as the original train set. In Sec. 4.5, the semi-supervised training uses RIMES train images without transcriptions; in Sec. 5.5 these are exactly the images whose text/style pairs are later replicated for CER-train. The model therefore sees the target images during adaptation (albeit without paired text labels), and the reported gain from 9.99% to 6.61% in Table 9 may be attributable to exposure to the adaptation images rather than to a transferable ability to generate new content in RIMES style. The authors should evaluate on a disjoint split of the target dataset, report a RIMES out-of-vocabulary metric, or otherwise separate the images used for adaptation from those used for CER-train. The qualitative failure on 'cartes' in Sec. 5.6 is consistent with the need for such a test.
  2. [Sec. 5.7, Table 9 vs Sec. 5.5, Table 5a] There is an internal inconsistency in the reported RIMES CER-train values for the same configuration. Table 9 lists 'Ours (IAM only)' as 10.12 at wgs=2 and 9.99 at wgs=5, while the TS column of Table 5a, which corresponds to the same setup (DM trained only on IAM, CFG with puncond=0.1, TS style inclusion), lists 9.58 and 9.86. Since Table 9 uses the reported IAM-only result as the baseline against which the semi-supervised improvement is measured, the authors must reconcile these numbers and clarify which model and MAE training set produced each entry.
  3. [Sec. 4.4] The claim that the MAE style embedding captures calligraphic style is not directly tested. The paper assumes that the unmasking objective 'pushes the model to encode information about the strokes' and that reconstruction forces the encoder to store style, but no writer-retrieval, writer-identification, or style-similarity metric is reported. The unseen-writer capability, the semi-supervised RIMES gains, and the comparison to DiffusionPen all depend on the averaged MAE embedding actually conveying per-writer style rather than mostly content. I recommend adding a quantitative style-fidelity evaluation, e.g., writer retrieval accuracy on held-out writers or a style-consistency measure on generated images.
  4. [Tables 3-6, 9] All results appear to come from single runs without error bars or multiple seeds. Many of the comparisons that support the paper's claims differ by less than one CER point (for example, several entries in Tables 5 and 6 differ by 0.1-0.5 percentage points), so the reader cannot assess whether these differences are significant. At minimum, the key semi-supervised comparison in Table 9 should be repeated with multiple seeds and reported with mean and standard deviation.
minor comments (5)
  1. [Sec. 5.2] The text refers to an 'HRT model' in the paragraph defining Diff-IV; this should be 'HTR model'.
  2. [Sec. 2.2] The phrase 'genration of online and offline handwriting' contains a typo; it should read 'generation'.
  3. [Eq. (34)] The notation SE is overloaded: SE denotes the encoder in Eq. (33) and the writer embedding function in Eq. (34). Rename one of these, e.g., use a separate symbol for the averaged writer embedding.
  4. [Table 5] The table caption states that MAEs were trained on different datasets and refers to a column 'MAE DS', but Table 5a has no such column. Please clarify which MAE training set was used for Table 5a or restructure the table.
  5. [Sec. 4.5] The sentence 'we replace the sequence of embedding vectors for the text conditioning with the same mask tokens as for classifier-free guidance' is vague; it should specify whether the mask token is appended to or substituted for the text sequence and how this interacts with the timestep embedding in the TS variant.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: conditionings are separate from evaluation targets, metrics are downstream HTR CER on held-out test sets, and there are no load-bearing self-citations.

full rationale

The paper's derivation chain is self-contained: the latent diffusion model is conditioned on a transcription string, a timestep embedding, and a style embedding computed by an MAE from separate example images; none of these conditioning inputs contains the evaluation targets. The metrics are external downstream HTR character error rates on the original IAM and RIMES test sets, with test-set writers disjoint from training writers as stated in Sec. 5.1. The semi-supervised RIMES experiments train the diffusion model on unlabeled RIMES train images and then measure CER-train, which evaluates an HTR model trained on generated images and tested on the original test set; while the adaptation and generation prompts share the RIMES train split, the reported CER is not a prediction of the adaptation-set labels, so no equation reduces to its own inputs. The paper contains no self-citations, and the MAE style-fidelity assumption in Sec. 4.4 is an empirical assumption rather than a circular definition. Guidance-scale selection is a hyperparameter search that could raise soundness or multiple-comparison concerns, but it is not a circularity of derivation. Overall, the central claims do not reduce by construction to the inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard diffusion machinery plus several domain-level modeling choices. The most consequential unverified choices are the MAE style-embedding assumption and the use of HTR CER as the sole quality proxy. No new physical entities are introduced; the style embedding is a learned representation, not a postulated entity.

free parameters (5)
  • guidance scale wgs = 2 (semi-supervised) / 5 (IAM-only)
    Grid searched over {2,3,4,5,6,7} in Tables 3-6; the headline comparison in Table 9 uses the best value per scenario with no held-out validation split.
  • number of style examples K = 10
    Sec. 5.3: embeddings averaged over 10 random example images per writer; chosen for practical availability, not optimized or varied.
  • MAE masking ratio = 0.75
    Sec. 5.3: 75% of 8x8 patches randomly masked during style encoder training; inherited from [76] setup.
  • conditioning drop probability puncond = 0.1
    Sec. 5.3: writer and text conditioning independently dropped with p=0.1 for classifier-free guidance; single choice, not swept.
  • style embedding sampling repeats = 100
    Sec. 5.3: 100 random writer embeddings per writer per diffusion training iteration used as augmentation; chosen without ablation.
assumptions (5)
  • domain assumption An MAE trained with unmasking alone encodes calligraphic style in averaged patch embeddings, not just content.
    Sec. 4.4 states the unmasking task is assumed to push stroke information into embeddings and reconstruction is assumed to force style encoding; no style-similarity measurement is provided.
  • domain assumption CER-train on a single HTR model is a valid proxy for generated-image quality and domain suitability.
    Sec. 5.2 uses CER-train and Diff-IV as primary metrics; no human evaluation or writer-style fidelity metric is reported.
  • domain assumption Writer identity for the new dataset is recoverable from page-level grouping.
    Sec. 5.1 assumes words from one page are written by one person; experiments use ground-truth RIMES writer labels instead of testing the inference procedure.
  • domain assumption The pre-trained Stable Diffusion VAE latent space preserves handwriting-relevant content and style after 64x256 preprocessing.
    Sec. 4.2 loads a frozen pre-trained VAE from stable-diffusion-v1-5 following Wordstylist; no handwriting-specific validation of the latent space is given.
  • standard math Standard DDPM training and sampling equations (forward process, epsilon prediction, classifier-free guidance) are accepted.
    Section 3 restates Ho et al. [30] and Ho and Salimans [31]; the paper contributes no new derivation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Semi-Supervised Adaptation of Diffusion Models for Handwritten Text Generation." pith.science (2026). https://pith.science/paper/NURFG7AE

@misc{pith2026241215853,
  author       = {Pith},
  title        = {Pith review of: Semi-Supervised Adaptation of Diffusion Models for Handwritten Text Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NURFG7AE}},
  note         = {Machine review of arXiv:2412.15853}
}
read the original abstract

The generation of images of realistic looking, readable handwritten text is a challenging task which is referred to as handwritten text generation (HTG). Given a string and examples from a writer, the goal is to synthesize an image depicting the correctly spelled word in handwriting with the calligraphic style of the desired writer. An important application of HTG is the generation of training images in order to adapt downstream models for new data sets. With their success in natural image generation, diffusion models (DMs) have become the state-of-the-art approach in HTG. In this work, we present an extension of a latent DM for HTG to enable generation of writing styles not seen during training by learning style conditioning with a masked auto encoder. Our proposed content encoder allows for different ways of conditioning the DM on textual and calligraphic features. Additionally, we employ classifier-free guidance and explore the influence on the quality of the generated training images. For adapting the model to a new unlabeled data set, we propose a semi-supervised training scheme. We evaluate our approach on the IAM-database and use the RIMES-database to examine the generation of data not seen during training achieving improvements in this particularly promising application of DMs for HTG.

Figures

Figures reproduced from arXiv: 2412.15853 by the authors.

Figure 1
Figure 1. Overview of the proposed system during training given an input image [PITH_FULL_IMAGE:figures/full_fig_p013_1.png] view at source ↗
Figure 2
Figure 2. Overview of the proposed system when sampling an image given the [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. Computation of the content embedding C = CE(y, sw) and the timestep embedding temb = CT (t, sw) for different choices of incorporating the style em￾bedding. PE and LPE denote positional encodings and learned positional em￾beddings, respectively. Operands are + for elementwise addition, L for concate￾nation of two vectors and ∥ for appending a vector to the sequence. receives the string y = (y1, . . . , yk), k ≤ l of… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Dual Orthogonal Guidance for Robust Diffusion-based Handwritten Text Generation

    cs.CV 2025-08 unverdicted novelty 6.0 of 10

    A new guidance method for diffusion-based text generation subtracts the orthogonal component of a negative prompt from a positive prompt to reduce artifacts and increase style variation.

Reference graph

Works this paper leans on

89 extracted references · 47 canonical work pages · cited by 1 Pith paper

  1. [1]

    In: Proc

    Aksan, E., Pece, F., Hilliges, O.: DeepWriting: Making Digital Ink Editable via Deep Generative Modeling. In: Proc. CHI Conference on Human Factors in Computing Systems. Montreal, Canada (2018). https://doi.org/10.1145/3173574. 3173779

    doi:10.1145/3173574 2018

  2. [2]

    Alonso, E., Moysset, B., Messina, R.: Adversarial Generation of Handwritten Text Images Conditioned on Sequences. In: Int. Conf. on Document Analysis and Recog- nition. pp. 481–486. Sydney, Australia (2019). https://doi.org/10.1109/ICDAR. 2019.00083

    arXiv 2019

  3. [3]

    arXiv: abs/1607.06450 (2016)

    Ba, J.L., Kiros, J.R., Hinton, G.E.: Layer Normalization. arXiv: abs/1607.06450 (2016). https://doi.org/10.48550/arXiv.1607.06450

    arXiv doi:10.48550/arxiv.1607.06450 2016

  4. [4]

    In: IEEE/CVF Int

    Bhunia, A., Khan, S., Cholakkal, H., Anwer, R., Khan, F., Shah, M.: Handwriting Transformers. In: IEEE/CVF Int. Conf. on Computer Vision. pp. 1066–1074. Los Alamitos, CA, USA (2021). https://doi.org/10.1109/ICCV48922.2021.00112

    arXiv 2021

  5. [5]

    Springer International Publishing (2024)

    Bishop, C.M., Bishop, H.: Deep Learning: Foundations and Concepts. Springer International Publishing (2024). https://doi.org/10.1007/978-3-031-45468-4

    doi:10.1007/978-3-031-45468-4 2024

  6. [6]

    Chang, C.C., Perera, L.P.G., Khudanpur, S.: Crosslingual Handwritten Text Generation Using GANs. In: Int. Conf. on Document Analysis and Recognition Workshops. pp. 285–301. San José, CA, USA (2023). https://doi.org/10.1007/ 978-3-031-41501-2_20

    2023

  7. [7]

    In: Advances in Neural Information Processing Systems

    Chung, J., Kastner, K., Dinh, L., Goel, K., Courville, A., Bengio, Y.: A Recurrent Latent Variable Model for Sequential Data. In: Advances in Neural Information Processing Systems. Montréal, Canada (2015)

    2015

  8. [8]

    In: European Conf

    Dai, G., Zhang, Y., Ke, Q., Guo, Q., Huang, S.: One-Shot Diffusion Mimicker for Handwritten Text Generation. In: European Conf. on Computer Vision. pp. 410–427. Milan, Italy (2024). https://doi.org/10.1007/978-3-031-73636-0_24

    work page doi:10.1007/978-3-031-73636-0_24 2024

Show all 89 references
  1. [9]

    In: British Machine Vision Conference

    Davis, B.L., Tensmeyer, C., Price, B.L., Wigington, C., Morse, B.S., Jain, R.: Text and Style Conditioned GAN for Generation of Offline Handwriting Lines. In: British Machine Vision Conference. Virtual Conference (2020)

    2020

  2. [10]

    In: IEEE/CVF Conf

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: ImageNet: A large- scale hierarchical image database. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 248–255. Miami, FL, USA (2009). https://doi.org/10. 1109/CVPR.2009.5206848

    2009

  3. [11]

    In: Advances in Neural Information Processing Systems

    Dhariwal, P., Nichol, A.: Diffusion Models Beat GANs on Image Synthesis. In: Advances in Neural Information Processing Systems. pp. 8780–8794. Virtual Con- ference (2021)

    2021

  4. [12]

    Ding, H., Luan, B., Gui, D., Chen, K., Huo, Q.: Improving Handwritten OCR with Training Samples Generated by Glyph Conditional Denoising Diffusion Probabilis- tic Model. In: Int. Conf. on Document Analysis and Recognition. pp. 20–37. San José, CA, USA (2023). https://doi.org/1...

    2023 doi

  5. [13]

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In: Int. Conf. on Learning Repres...

    2021

  6. [14]

    The Visual Computer (2024)

    Elanwar, R., Betke, M.: Generative Adversarial Networks for Handwriting Im- age Generation: A Review. The Visual Computer (2024). https://doi.org/10.1007/ s00371-024-03534-9

    2024

  7. [15]

    In: IEEE/CVF Conf

    Esser, P., Rombach, R., Ommer, B.: Taming Transformers for High-Resolution Image Synthesis. In: IEEE/CVF Conf. on Computer Vision and Pattern Recog- 32 K. Brandenbusch nition. pp. 12868–12878. Virtual Conference (2021). https://doi.org/10.1109/ CVPR46437.2021.01268

    2021

  8. [16]

    In: IEEE/CVF Conf

    Fogel, S., Averbuch-Elor, H., Cohen, S., Mazor, S., Litman, R.: ScrabbleGAN: Semi-Supervised Varying Length Handwritten Text Generation. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 4323–4332. Seattle, WA, USA (2020). https://doi.org/10.1109/CVPR42600.2020.00438

    2020

  9. [17]

    In: Proc

    Gan, J., Wang, W.: HiGAN: Handwriting Imitation Conditioned on Arbitrary- Length Texts and Disentangled Styles. In: Proc. AAAI Conf. on Artificial Intel- ligence. pp. 7484–7492. Virtual Conference (2021). https://doi.org/10.1609/aaai. v35i9.16917

    2021 doi

  10. [18]

    ACM Trans

    Gan, J., Wang, W., Leng, J., Gao, X.: HiGAN+: Handwriting Imitation GAN with Disentangled Representations. ACM Trans. Graph.42(1), 1–17 (2022). https: //doi.org/10.1145/3550070

    2022 doi

  11. [19]

    Ganin, Y., Kulkarni, T., Babuschkin, I., Eslami, S.M.A., Vinyals, O.: Synthesiz- ing Programs for Images using Reinforced Adversarial Learning. In: Int. Conf. on Machine Learning. pp. 1666–1675. Stockholm, Sweden (2018)

    2018

  12. [20]

    In: Advances in Neural Information Processing Systems

    Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.: Generative Adversarial Nets. In: Advances in Neural Information Processing Systems. Montréal, Canada (2014)

    2014

  13. [21]

    arXiv: abs/1308.0850 (2013)

    Graves, A.: Generating Sequences With Recurrent Neural Networks. arXiv: abs/1308.0850 (2013). https://doi.org/10.48550/arXiv.1308.0850

    2013 doi arXiv

  14. [22]

    Graves, A., Fernández, S., Gomez, F., Schmidhuber, J.: Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with Recurrent Neural Net- works. In: Int. Conf. on Machine Learning. pp. 369–376. Pittsburgh, PA, USA (2006). https://doi.org/10.1145/1143844.1143891

    2006

  15. [23]

    In: Int.Conf.onDocumentAnalysisandRecognition.pp.1398–1402.Barcelona,Spain (2009)

    Grosicki, E., Abed, H.E.: ICDAR 2009 Handwriting Recognition Competition. In: Int.Conf.onDocumentAnalysisandRecognition.pp.1398–1402.Barcelona,Spain (2009). https://doi.org/10.1109/ICDAR.2009.184

    2009 doi

  16. [24]

    Guan, M., Ding, H., Chen, K., Huo, Q.: Improving Handwritten OCR with Aug- mented Text Line Images Synthesized from Online Handwriting Samples by Style- Conditioned GAN. In: Int. Conf. on Frontiers in Handwriting Recognition. pp. 151–

  17. [25]

    Gui, D., Chen, K., Ding, H., Huo, Q.: Zero-shot Generation of Training Data with Denoising Diffusion Probabilistic Model for Handwritten Chinese Character Recognition. In: Int. Conf. on Document Analysis and Recognition. pp. 348–365. San José, CA, USA (2023). https://doi.org/1...

    2023 doi

  18. [26]

    ACM Trans

    Haines, T.S.F., Mac Aodha, O., Brostow, G.J.: My Text in Your Handwriting. ACM Trans. on Graphics35(3), 1–18 (2016). https://doi.org/10.1145/2886099

    2016 doi

  19. [27]

    In: IEEE/CVF Conf

    He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.: Masked Autoencoders Are Scalable Vision Learners. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 15979–15988. New Orleans, LA, USA (2022). https://doi.org/10. 1109/CVPR52688.2022.01553

    2022

  20. [28]

    In: IEEE/CVF Conf

    He, K., Zhang, X., Ren, S., Sun, J.: Deep Residual Learning for Image Recognition. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 770–778. Las Vegas, NV, USA (2016)

    2016

  21. [29]

    In: Advances in Neural Information Processing Systems

    Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., Hochreiter, S.: GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilib- rium. In: Advances in Neural Information Processing Systems. Long Beach, CA, USA (2017)

    2017

  22. [30]

    In: Advances in Neural Information Processing Systems

    Ho, J., Jain, A., Abbeel, P.: Denoising Diffusion Probabilistic Models. In: Advances in Neural Information Processing Systems. Virtual Conference (2020) Semi-Supervised Adaptation of Diffusion Models for HTG 33

    2020

  23. [31]

    In: NeurIPS Workshop on Deep Generative Models and Downstream Applications (2021)

    Ho, J., Salimans, T.: Classifier-Free Diffusion Guidance. In: NeurIPS Workshop on Deep Generative Models and Downstream Applications (2021). https://doi.org/ 10.48550/arXiv.2207.12598

    2021 doi arXiv

  24. [32]

    Neural Computing 9(8), 1735–1780 (1997)

    Hochreiter, S., Schmidhuber, J.: Long Short-Term Memory. Neural Computing 9(8), 1735–1780 (1997). https://doi.org/10.1162/neco.1997.9.8.1735

    1997 doi

  25. [33]

    Ingle, R.R., Fujii, Y., Deselaers, T., Baccash, J., Popat, A.C.: A Scalable Hand- written Text Recognition System. In: Int. Conf. on Document Analysis and Recog- nition. pp. 17–24. Sydney, Australia (2019). https://doi.org/10.1109/ICDAR.2019. 00013

    2019 doi

  26. [34]

    In: IEEE/CVF Conf

    Isola, P., Zhu, J.Y., Zhou, T., Efros, A.A.: Image-To-Image Translation With Con- ditional Adversarial Networks. In: IEEE/CVF Conf. on Computer Vision and Pat- tern Recognition. pp. 5967–5976. Honolulu, HI, USA (2017). https://doi.org/10. 1109/CVPR.2017.632

    2017

  27. [35]

    arXiv: abs/1907.11845 (2019)

    Ji, B., Chen, T.: Generative Adversarial Network for Handwritten Text. arXiv: abs/1907.11845 (2019). https://doi.org/10.48550/arXiv.1907.11845

    2019 doi arXiv

  28. [36]

    Kang, L., Riba, P., Rusiñol, M., Fornés, A., Villegas, M.: Distilling Content from Style for Handwritten Word Recognition. In: Int. Conf. on Frontiers in Handwrit- ing Recognition. pp. 139–144. Virtual Conference (2020). https://doi.org/10.1109/ ICFHR2020.2020.00035

    2020

  29. [37]

    IEEE Trans

    Kang, L., Riba, P., Rusiñol, M., Fornés, A., Villegas, M.: Content and Style Aware Generation of Text-Line Images for Handwriting Recognition. IEEE Trans. on Pattern Analysis and Machine Intelligence44(12), 8846–8860 (2022). https://doi. org/10.1109/TPAMI.2021.3122572

    2022

  30. [38]

    In: Euro- pean Conf

    Kang, L., Riba, P., Wang, Y., Rusiñol, M., Fornés, A., Villegas, M.: GANwriting: Content-Conditioned Generation of Styled Handwritten Word Images. In: Euro- pean Conf. on Computer Vision. pp. 273–289. Glasgow, United Kingdom (2020). https://doi.org/10.1007/978-3-030-58592-1_17

    2020 doi

  31. [39]

    In: Proc

    Kang, L., Rusinol, M., Fornes, A., Riba, P., Villegas, M.: Unsupervised Writer Adaptation for Synthetic-to-Real Handwritten Word Recognition. In: Proc. of the IEEE/CVF Winter Conf. on Applications of Computer Vision. pp. 3491–3500. Snowmass Village, CO, USA (2020). https://doi...

    2020

  32. [40]

    In: Proc

    Kang, L., Toledo, J.I., Riba, P., Villegas, M., Fornés, A., Rusiñol, M.: Convolve, At- tend and Spell: An Attention-based Sequence-to-Sequence Model for Handwritten Word Recognition. In: Proc. German Conf. on Pattern Recognition. pp. 459–472. Stuttgart, Germany (2018). https:/...

    2018 doi

  33. [41]

    Konidaris,T.,Gatos,B.,Ntzios,K.,Pratikakis,I.,Theodoridis,S.,Perantonis,S.J.: Keyword-Guided Word Spotting in Historical Printed Documents Using Synthetic Data and User Feedback. Int. Journal on Document Analysis and Recognition 9(2–4), 167–177 (Mar 2007). https://doi.org/10.1...

    2007 doi

  34. [42]

    arXiv: abs/1608.04224 (2016)

    Krishnan, P., Jawahar, C.V.: Generating Synthetic Data for Text Recognition. arXiv: abs/1608.04224 (2016). https://doi.org/10.48550/arXiv.1608.04224

    2016 doi arXiv

  35. [43]

    Krishnan, P., Jawahar, C.V.: HWNet v2: an efficient word image representation for handwritten documents. Int. Journal on Document Analysis and Recognition 22(4), 387–405 (2019). https://doi.org/10.1007/s10032-019-00336-x

    2019 doi

  36. [44]

    arXiv: abs/2106.08385 (2021)

    Krishnan, P., Kovvuri, R., Pang, G., Vassilev, B., Hassner, T.: TextStyleBrush: Transfer of Text Aesthetics from a Single Example. arXiv: abs/2106.08385 (2021). https://doi.org/10.48550/ARXIV.2106.08385, 2023 pusblished in PAMI

    2021 doi arXiv

  37. [45]

    Lee, J., Lee, Y., Kim, J., Kosiorek, A., Choi, S., Teh, Y.W.: Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks. In: Int. Conf. on Machine Learning. pp. 3744–3753. Long Beach, CA, USA (2019) 34 K. Brandenbusch

    2019

  38. [46]

    In: Proc

    Li, M., Lv, T., Chen, J., Cui, L., Lu, Y., Florencio, D., Zhang, C., Li, Z., Wei, F.: TrOCR: Transformer-based Optical Character Recognition with Pre-trained Mod- els. In: Proc. AAAI Conf. on Artificial Intelligence. pp. 13094–13102. Washington DC, USA (2023)

    2023

  39. [47]

    Pattern Recog- nition 40(7), 2097–2109 (2007)

    Lin, Z., Wan, L.: Style-preserving English handwriting synthesis. Pattern Recog- nition 40(7), 2097–2109 (2007). https://doi.org/10.1016/j.patcog.2006.11.024

    2007 doi

  40. [48]

    IEEE Signal Processing Letters 28, 1838–1842 (2021)

    Liu, X., Meng, G., Xiang, S., Pan, C.: Handwritten Text Generation via Disen- tangled Representations. IEEE Signal Processing Letters 28, 1838–1842 (2021). https://doi.org/10.1109/LSP.2021.3109541

    2021

  41. [49]

    Loshchilov, I., Hutter, F.: Decoupled Weight Decay Regularization. In: Int. Conf. on Learning Representations. New Orleans, LA, USA (2019)

    2019

  42. [50]

    In: Advances in Neural Information Processing Systems

    Lu, C., Zhou, Y., Bao, F., Chen, J., Chongxuan, L.I., Zhu, J.: DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps. In: Advances in Neural Information Processing Systems. pp. 5775–5787. New Orleans, LA, USA (2022)

    2022

  43. [51]

    arXiv: abs/2211.01095 (2022)

    Lu, C., Zhou, Y., Bao, F., Chen, J., Li, C., Zhu, J.: DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models. arXiv: abs/2211.01095 (2022). https://doi.org/10.48550/arXiv.2211.01095

    2022 doi arXiv

  44. [52]

    arXiv: abs/2011.06704 (2020)

    Luhman, T., Luhman, E.: Diffusion models for Handwriting Generation. arXiv: abs/2011.06704 (2020). https://doi.org/10.48550/arXiv.2011.06704

    2020 doi arXiv

  45. [53]

    IEEE Trans

    Luo, C., Zhu, Y., Jin, L., Li, Z., Peng, D.: SLOGAN: Handwriting Style Synthesis for Arbitrary-Length and Out-of-Vocabulary Text. IEEE Trans. on Neural Net- works and Learning Systems34(11), 8503–8515 (2023). https://doi.org/10.1109/ TNNLS.2022.3151477

    2023

  46. [54]

    Marti,U.V.,Bunke,H.:TheIAM-database:anEnglishsentencedatabaseforoffline handwriting recognition. Int. Journal on Document Analysis and Recognition5(1), 39–46 (2002). https://doi.org/10.1007/s100320200071

    2002 doi

  47. [55]

    In: Lladós, J., Lopresti, D., Uchida, S

    Mattick, A., Mayr, M., Seuret, M., Maier, A., Christlein, V.: SmartPatch: Im- proving Handwritten Word Imitation with Patch Discriminators. In: Lladós, J., Lopresti, D., Uchida, S. (eds.) Int. Conf. on Document Analysis and Recog- nition. pp. 268–283. Lausanne, Switzerland (20...

    2021

  48. [56]

    arXiv: abs/2409.00786 (2024)

    Mayr, M., Dreier, M., Kordon, F., Seuret, M., Zöllner, J., Wu, F., Maier, A., Christlein,V.:Zero-ShotParagraph-levelHandwritingImitationwithLatentDiffu- sion Models. arXiv: abs/2409.00786 (2024). https://doi.org/10.48550/arXiv.2409. 00786

    2024 doi arXiv

  49. [57]

    In: European Conf

    Mayr, M., Stumpf, M., Nicolaou, A., Seuret, M., Maier, A., Christlein, V.: Spatio- Temporal Handwriting Imitation. In: European Conf. on Computer Vision Work- shops. pp. 528–543. Glasgow, United Kingdom (2020). https://doi.org/10.1007/ 978-3-030-68238-5_38

    2020

  50. [58]

    Michael, J., Labahn, R., Grüning, T., Zöllner, J.: Evaluating Sequence-to-Sequence ModelsforHandwrittenTextRecognition.In:Int.Conf.onDocumentAnalysisand Recognition. pp. 1286–1293. Sydney, Australia (2019). https://doi.org/10.1109/ ICDAR.2019.00208

    2019

  51. [59]

    arXiv: abs/1411.1784 (2014)

    Mirza, M., Osindero, S.: Conditional Generative Adversarial Nets. arXiv: abs/1411.1784 (2014). https://doi.org/10.48550/arXiv.1411.1784

    2014 doi arXiv

  52. [60]

    Miyato, T., Koyama, M.: cGANs with Projection Discriminator. In: Int. Conf. on Learning Representations. Vancouver, Canada (2018)

    2018

  53. [61]

    Nichol, A.Q., Dhariwal, P.: Improved Denoising Diffusion Probabilistic Model. In: Int. Conf. on Machine Learning. pp. 8162–8171. Virtual Conference (2021) Semi-Supervised Adaptation of Diffusion Models for HTG 35

    2021

  54. [62]

    Nichol, A.Q., Dhariwal, P., Ramesh, A., Shyam, P., Mishkin, P., Mcgrew, B., Sutskever, I., Chen, M.: GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Models. In: Int. Conf. on Machine Learning. pp. 16784–16804. Baltimore, MD, USA (2022)

    2022

  55. [63]

    Nikolaidou, K., Retsinas, G., Christlein, V., Seuret, M., Sfikas, G., Smith, E.B., Mokayed, H., Liwicki, M.: Wordstylist: Styled Verbatim Handwritten Text Gen- eration with Latent Diffusion Models. In: Int. Conf. on Document Analysis and Recognition. pp. 384–401. San José, CA,...

    2023

  56. [64]

    In: European Conf

    Nikolaidou, K., Retsinas, G., Sfikas, G., Liwicki, M.: DiffusionPen: Towards Controlling the Style of Handwritten Text Generation. In: European Conf. on Computer Vision. pp. 417–434. Milan, Italy (2024). https://doi.org/10.1007/ 978-3-031-73013-9_24

    2024

  57. [65]

    In: Advances in Neural Information Processing Systems

    van den Oord, A., Vinyals, O., kavukcuoglu, k.: Neural Discrete Representation Learning. In: Advances in Neural Information Processing Systems. Long Beach, CA, USA (2017)

    2017

  58. [66]

    In: IEEE/CVF Conf

    Pippi,V., Cascianelli,S., Cucchiara, R.:Handwritten Text GenerationFromVisual Archetypes. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 22458–22467. Vancouver, Canada (2023)

    2023

  59. [67]

    Retsinas,G.,Sfikas,G.,Gatos,B.,Nikou,C.:BestPracticesforaHandwrittenText Recognition System. In: Int. Workshop on Document Analysis Systems. pp. 247–

  60. [68]

    Riaz, N., Saifullah, S., Agne, S., Dengel, A., Ahmed, S.: StylusAI: Stylistic Adaptation for Robust German Handwritten Text Generation. In: Int. Conf. on Document Analysis and Recognition. pp. 429–444. Athens, Greece (2024). https://doi.org/10.1007/978-3-031-70536-6_26

    2024 doi

  61. [69]

    In: IEEE/CVF Conf

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-Resolution Image Synthesis With Latent Diffusion Models. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 10684–10695. New Orleans, LA, USA (2022)

    2022

  62. [70]

    Ronneberger, O., Fischer, P., Brox, T.: U-Net: Convolutional Networks for Biomed- ical Image Segmentation. In: Int. Conf. on Medical Image Computing and Computer-Assisted Intervention. pp. 234–241. Munich, Germany (2015). https: //doi.org/10.1007/978-3-319-24574-4_28

    2015 doi

  63. [71]

    In: IEEE/CVF Conf

    Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: MobileNetV2: In- verted Residuals and Linear Bottlenecks. In: IEEE/CVF Conf. on Computer Vi- sion and Pattern Recognition. pp. 4510–4520. Salt Lake City, UT, USA (2018). https://doi.org/10.1109/CVPR.2018.00474

    2018

  64. [72]

    Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., Ganguli, S.: Deep Unsuper- vised Learning using Nonequilibrium Thermodynamics. In: Int. Conf. on Machine Learning. pp. 2256–2265. Lille, France (2015)

    2015

  65. [73]

    Song, J., Meng, C., Ermon, S.: Denoising Diffusion Implicit Models. In: Int. Conf. on Learning Representations. Vienna, Austria (2021)

    2021

  66. [74]

    In: Advances in Neural Information Processing Systems

    Song, Y., Ermon, S.: Generative Modeling by Estimating Gradients of the Data Distribution. In: Advances in Neural Information Processing Systems. Vancouver, Canada (2019)

    2019

  67. [75]

    Song, Y., Sohl-Dickstein, J., Kingma, D.P., Kumar, A., Ermon, S., Poole, B.: Score-Based Generative Modeling through Stochastic Differential Equations. In: Int. Conf. on Learning Representations. Vienna, Austria (2021)

    2021

  68. [76]

    Brandenbusch invariant autoencoder for text recognition and document enhancement

    Souibgui, M.A., Biswas, S., Mafla, A., Biten, A.F., Fornés, A., Kessentini, Y., Lladós, J., Gomez, L., Karatzas, D.: Text-DIAE: a self-supervised degradation 36 K. Brandenbusch invariant autoencoder for text recognition and document enhancement. In: Proc. AAAI Conf. on Artific...

    2023 doi

  69. [77]

    In: IEEE/CVF Conf

    Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., Wojna, Z.: Rethinking the In- ception Architecture for Computer Vision. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 2818–2826. Las Vegas, NV, USA (2016). https://doi.org/10.1109/CVPR.2016.308

    2016 doi

  70. [78]

    Pattern Recognition 42(12), 3365–3373 (2009)

    Thomas, A.O., Rusu, A., Govindaraju, V.: Synthetic handwritten CAPTCHAs. Pattern Recognition 42(12), 3365–3373 (2009). https://doi.org/10.1016/j.patcog. 2008.12.018

    2009 doi

  71. [79]

    arXiv: abs/2402.10798 (2024)

    Vanherle, B., Pippi, V., Cascianelli, S., Michiels, N., Van Reeth, F., Cucchiara, R.: VATr++: Choose Your Words Wisely for Handwritten Text Generation. arXiv: abs/2402.10798 (2024). https://doi.org/10.48550/arXiv.2402.10798

    2024 doi arXiv

  72. [80]

    In: Advances in Neural Information Processing Systems

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, L., Polosukhin, I.: Attention is All you Need. In: Advances in Neural Information Processing Systems. Long Beach, CA, USA (2017)

    2017

  73. [81]

    Wang, J., Wu, C., Xu, Y.Q., Shum, H.Y.: Combining Shape and Physical Models for Online Cursive Handwriting Synthesis. Int. Journal on Document Analysis and Recognition 7(4), 219–227 (2005). https://doi.org/10.1007/s10032-004-0131-6

    2005 doi

  74. [82]

    Wang, Y., Wang, H., Sun, S., Wei, H.: An Approach Based on Transformer and Deformable Convolution for Realistic Handwriting Samples Generation. In: Int. Conf. on Pattern Recognition. pp. 1457–1463. Montréal, Canada (2022). https: //doi.org/10.1109/ICPR56361.2022.9956551

    2022

  75. [83]

    In: ACM Int

    Zdenek, J., Nakayama, H.: JokerGAN: Memory-Efficient Model for Handwritten Text Generation with Text Line Awareness. In: ACM Int. Conf. on Multimedia. pp. 5655–5663. Virtual Conference (2021). https://doi.org/10.1145/3474085.3475713

    2021

  76. [84]

    Zdenek, J., Nakayama, H.: Handwritten Text Generation with Character- Specific Encoding for Style Imitation. In: Int. Conf. on Document Analysis and Recognition. pp. 313–329. San José, CA, USA (2023). https://doi.org/10.1007/ 978-3-031-41679-8_18

    2023

  77. [85]

    Zhang, Q., Chen, Y.: Fast Sampling of Diffusion Models with Exponential Integra- tor. In: Int. Conf. on Learning Representations. Kigali, Rwanda (2023)

    2023

  78. [86]

    In: IEEE/CVF Conf

    Zhang, R., Isola, P., Efros, A.A., Shechtman, E., Wang, O.: The Unreasonable Effectiveness of Deep Features as a Perceptual Metric. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 586–595. Salt Lake City, UT, USA (2018). https://doi.org/10.1109/CVPR.2018.00068

    2018

  79. [87]

    In: Advances in Neural Infor- mation Processing Systems

    Zhao, W., Bai, L., Rao, Y., Zhou, J., Lu, J.: UniPC: A Unified Predictor-Corrector Framework for Fast Sampling of Diffusion Models. In: Advances in Neural Infor- mation Processing Systems. New Orleans, LA, USA (2023)

    2023

  80. [88]

    In: IEEE/CVF Conf

    Zhu, Y., Li, Z., Wang, T., He, M., Yao, C.: Conditional Text Image Generation With Diffusion Models. In: IEEE/CVF Conf. on Computer Vision and Pattern Recognition. pp. 14235–14245. Vancouver, Canada (2023)

    2023

  81. [259]

    https://doi.org/10.1007/978-3-031-06555-2_17

    La Rochelle, France (2022). https://doi.org/10.1007/978-3-031-06555-2_17

    2022 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.