Pith. sign in

REVIEW 5 major objections 6 minor 30 references

Sign Language Recognition Using Original and Synthetic Depth Image Based Point Cloud Data Models

T0 review · 5 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read This paper tests whether depth images generated from ordinary RGB video can replace real depth-camera data for sign language recognition, reporting that point clouds from synthetic depth match or beat original depth in some models.

desk verdict Useful empirical extension of synthetic-depth SLR, but test-set-based epoch selection makes the headline accuracy gaps unreliable. read the letter →

arxiv 2608.09400 v1 pith:WF6KELX4 submitted 2026-08-10 cs.CV cs.AI

classification cs.CVcs.AI
keywords signlanguagerecognitionpointcloudclassificationsyntheticdepthimagesAnythingV2NetGestureMapLSTMimageconversion
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

Sign language recognition usually depends on RGB video, while depth cameras that give 3D point clouds are scarce and expensive. This paper asks whether depth images generated from ordinary RGB frames by the Depth Anything V2 network can stand in for real depth-camera data when building point-cloud classifiers with PointNet. Using three public sign-language datasets and three data representations (single frames, Point Gesture Maps, and LSTM sequences), it compares accuracies from original and synthetic depth point clouds. The paper reports that synthetic depth reaches acceptable accuracy in most models and, in two KArSL configurations, beats original depth: 86.81% vs 32.11% for Point Gesture Maps and 97.56% vs 95.19% for LSTM. If this holds, any RGB-only sign-language video could be converted into point-cloud training data without a depth sensor.

What carries the argument

The mechanism that carries the argument is the replacement of time-of-flight depth sensing with monocular depth estimation before the point-cloud stage: Depth Anything V2 turns RGB frames into synthetic depth maps, Open3D converts both original and synthetic depth into 3D point clouds, and PointNet classifies them. Three data models organize the point clouds: frame-based input of 512 sampled points, Point Gesture Maps that merge a gesture's frames along an axis and sample 6400 points, and LSTM sequences built from 25 or 30 frame-level PointNet features with proportional frame interpolation for short videos. The comparison runs all models against held-out test data and reports accuracy, recall, specificity, and F1.

What would settle it

Re-run the KArSL LSTM and PGM comparisons with model selection done on the validation folds only, then report the test accuracy of the chosen models; if the synthetic advantage (97.56% vs 95.19% and 86.81% vs 32.11%) shrinks or flips, the headline result is an artifact of selecting the best test epoch.

Watch

Extended reading notes

Core claim

The paper's central claim is that synthetic depth images produced from monocular RGB frames by Depth Anything V2 can support sign language recognition from point clouds at a level close to real depth data, and occasionally above it. The authors build point clouds from both original sensor depth and synthetic depth using Open3D, train PointNet classifiers on three datasets, and evaluate three data models: per-frame point clouds, Point Gesture Maps that stack a gesture's frames, and LSTM networks fed by PointNet-extracted features. On most configurations the original depth models score higher, but on the KArSL dataset the synthetic Point Gesture Map model reaches 86.81% accuracy versus 32.11% for the original, and the synthetic LSTM model reaches 97.56% versus 95.19%. The paper interprets these cases as evidence that synthetic depth can create a distinctive positive effect that is not present in the original data.

Load-bearing premise

The comparison assumes that choosing the epoch with the highest accuracy on the held-out test set for each model is a fair way to measure how well original and synthetic depth point clouds perform, and that this test-set peeking does not systematically favor one data source over the other.

Editorial extensions

If this is right

  • Synthetic depth can substitute for missing depth in point-cloud sign language recognition, so existing RGB-only sign language video could become usable for depth-based methods.
  • In at least one dataset, synthetic depth point clouds outperform original depth point clouds, indicating the depth source is not always the limiting factor.
  • Point Gesture Map models can preserve or even improve accuracy with synthetic depth when frame-based models degrade, so temporal aggregation changes how sensitive the classifier is to depth-source shifts.
  • The approach removes the need for a depth sensor at inference time, since depth is generated from the RGB stream.
  • Because synthetic depth is close to original depth on most configurations, it could support pretraining or data augmentation for larger sign language models.

Reading between the lines

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

  • The reported accuracies select the best epoch on the test set for each model, so the absolute numbers are optimistic; a fair original-versus-synthetic comparison would need fixed training schedules or validation-based early stopping.
  • The large KArSL PGM gap (86.81% synthetic vs 32.11% original) suggests the original depth stream in that dataset may contain noise or missing regions that synthetic depth accidentally smooths; inspecting the original depth sensor's failure modes would clarify whether the advantage is real signal or artifact.
  • The same RGB-to-depth pipeline could be tested on continuous sign language recognition or on hand-region-cropped point clouds, where removing the background and signer body from the raw clouds might erase the remaining gap to original depth.
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

5 major / 6 minor

Summary. This paper compares sign language recognition (SLR) using point clouds derived from original depth images versus synthetic depth images generated from RGB frames with the Depth Anything V2 network. Three datasets (Real-time ASL Fingerspelling, KArSL, AUTSL) are used, and three data models are evaluated: frame-based PointNet, Point Gesture Map (PGM), and LSTM on PointNet-extracted features. The central empirical claim is that original-depth point clouds generally perform better, but synthetic-depth point clouds reach acceptable accuracy in most models and in some cases outperform original-depth models, most notably KArSL LSTM (97.56% vs. 95.19%) and KArSL PGM (86.81% vs. 32.11%). The paper reports accuracy, recall, specificity, and F1 scores for each model and states that code and pretrained models are publicly available.

Significance. If the comparison were unbiased, this result would be practically significant: it would indicate that monocular depth estimation can substitute for dedicated depth sensors in SLR point-cloud pipelines, alleviating the scarcity of depth-labeled sign language data. The study covers three large public datasets, uses multiple PointNet-based architectures, and the authors provide code and pretrained models, which strengthens reproducibility. However, the evaluation protocol as described selects models by test-set accuracy, so the current numbers cannot be interpreted as unbiased generalization estimates; the significance of the claimed synthetic-versus-original differences hinges on a clean re-analysis.

major comments (5)
  1. [Section 4, first paragraph] The protocol states that 'all cross-validation models for all the epochs were tested using the test data' and that the epoch with the highest averaged test accuracy is selected. This makes every Table 6 entry an optimistic maximum over epochs and can bias the original-versus-synthetic comparison because epoch budgets differ between modalities (e.g., Dataset-B Frame: 240 original vs. 50 synthetic; Dataset-C LSTM: 100 original vs. 200 synthetic). Model selection must be based on the validation folds (or a separate validation set), with the test set used only once, and the chosen epoch should be reported.
  2. [Section 3.2, Dataset-B and Dataset-C LSTM preparation] LSTM input features are extracted from the 'best performing pretrained frame based PointNet,' and the text does not specify whether that frame model was selected using validation accuracy or test accuracy. Since the frame models are selected by the same test-set-peeking procedure, test information propagates into the LSTM representations, making the LSTM accuracies in Table 6 non-independent of the test set. The selection criterion for the feature extractor must be stated and, if it used the test set, the feature extraction must be redone with validation-based selection.
  3. [Table 6] The Dataset-C PGM (Synthetic Depth) row is missing entirely; Section 4 states only that this model achieved 'an insignificant accuracy' and omits its confusion matrix, while all other models have numeric entries. A numerical result, or an explicit explanation of why this condition is excluded, is necessary for the table to support the claimed cross-modality comparison.
  4. [Table 6, Dataset-B PGM rows] The gap between original-depth PGM (32.11%) and synthetic-depth PGM (86.81%) is so large that it suggests a pipeline-level problem rather than a genuine modality effect, especially because the original-depth PGM accuracy is far below the corresponding frame-based and LSTM accuracies. The paper does not analyze this case (e.g., point-cloud quality, label alignment, or the training curves in Figures 15 and 16), yet it is cited in Section 5 as evidence that synthetic data can be superior. This anomaly must be investigated before the claim can be accepted.
  5. [Section 4 and Table 6] No variance or confidence intervals are reported: all results are point estimates from a single test split, so small differences such as Dataset-B LSTM 95.19% vs. 97.56% cannot be assessed for significance. Per-fold results, standard deviations, or confidence intervals should be reported for every model to allow the reader to evaluate whether the observed gaps are meaningful.
minor comments (6)
  1. [Abstract] The word 'whileas' is a typo; it should be 'whereas'.
  2. [Section 1] 'perfomance' is a typo; it should be 'performance'.
  3. [Table 6 header] The column header 'Precision (Accuracy)' conflates two distinct metrics; the column contains accuracy values, so it should be labeled 'Accuracy', and the precision metric should be defined elsewhere if reported.
  4. [Figure 14 caption] 'Dataset-b' should be capitalized as 'Dataset-B' for consistency.
  5. [Section 3.3] The sentence 'In some training figures, a few very high spikes of error were omitted' should be disclosed in each affected figure caption or in the main text, because omitted spikes affect the reader's interpretation of the training curves.
  6. [References] Reference [30] is listed in the bibliography but is not cited in the text; either cite it where relevant or remove it.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the accuracy comparison is empirical, externally benchmarked, and does not reduce to fitted parameters or self-citations.

full rationale

The paper's central claim is an empirical comparison: original versus synthetic depth images converted to point clouds and classified with PointNet-based models. The synthetic depths come from the external Depth Anything V2 network [23], the point clouds are built with external Open3D [19], and classification accuracies are measured on held-out test partitions of three public datasets. No prediction is derived from a fitted parameter, and no uniqueness theorem is invoked. The only self-citation is [20], used to justify the Point Gesture Map stacking construction; this reuses a geometric data representation from the authors' prior work, but the reported accuracies are produced by newly trained models on external test data, so the result does not reduce to the cited paper. The most notable methodological concern is the Section 4 protocol that selects each model's best epoch using accuracy on the test set, which biases the reported accuracies and could differentially affect original versus synthetic pipelines; this is a correctness and generalization risk, not a circularity, because the reported numbers are still evaluated on test data rather than being identical to the selection criterion by construction. No self-definitional, fitted-input-called-prediction, ansatz-smuggled-via-citation, or renaming pattern is present.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new entities, particles, forces, or conserved quantities. The load-bearing assumptions are the validity of synthetic depth for point-cloud generation, the fairness of test-set-based epoch selection, and the adequacy of raw (unsegmented) point clouds for SLR.

free parameters (7)
  • depth_scale = 1.0 (Dataset-A), 500 (Dataset-B/C)
    Hand-selected parameter in Open3d that scales depth values before point cloud creation; directly changes point coordinates.
  • depth_trunc = 1000
    Hand-selected maximum depth in millimeters for point cloud creation.
  • sample_point_count = 512 (frame), 6400 (PGM)
    Number of points sampled per point cloud for PointNet input; chosen by the authors.
  • lstm_frame_count = 25 (Dataset-B), 30 (Dataset-C)
    Fixed number of frames per gesture for LSTM; chosen after examining average frame counts.
  • pgm_frame_merge = 49 (Dataset-A); all frames (Dataset-B/C)
    Number of frames merged into a Point Gesture Map; Dataset-A used 49 of about 500 frames.
  • pgm_dense_layers = 4096-2048-1024-512 with dropout 0.3
    Customized PointNet last layers for PGM models, determined by experiment.
  • lstm_dense_layers = LSTM 256, dense 2048, 1024, dropout 0.2
    Customized classification head for LSTM models, determined by experiment.
assumptions (4)
  • domain assumption Depth Anything V2 produces depth maps that can be converted into geometrically meaningful point clouds using Open3d intrinsics and depth-scale/trunc parameters.
    The entire synthetic branch depends on this unvalidated equivalence; the paper does not calibrate the synthetic depth scale to the Kinect cameras.
  • ad hoc to paper Selecting the best epoch by test-set accuracy does not bias the comparison between original and synthetic models.
    Section 4 describes test-set peeking; the paper implicitly assumes this affects both modalities equally.
  • domain assumption Point clouds containing the full scene, without hand/arm separation, are adequate inputs for SLR classification.
    Section 5 says raw data was used without arm/hand separation or processing.
  • domain assumption The frame interpolation method (blending previous and next frames) preserves the gesture semantics for LSTM training.
    Used for Dataset-B/C LSTM data when frames are between 14-25 or 20-30; no validation of this new synthetic frame content.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Sign Language Recognition Using Original and Synthetic Depth Image Based Point Cloud Data Models." pith.science (2026). https://pith.science/paper/WF6KELX4

@misc{pith2026260809400,
  author       = {Pith},
  title        = {Pith review of: Sign Language Recognition Using Original and Synthetic Depth Image Based Point Cloud Data Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WF6KELX4}},
  note         = {Machine review of arXiv:2608.09400}
}
read the original abstract

Research regarding the sign language recognition mostly relies on RGB images, whileas sign language datasets that provide depth images are limited. Point clouds obtained from depth images can be used for sign language recognition with neural networks like PointNet. In recent years, various neural networks are used for generating realistic depth images from monocular RGB images. In this work, synthetic depth images were created from RGB images using Depth Anything V2 network. For this purpose, three sign language datasets (Real-time ASL Fingerspelling, KArSL, AUTSL) which contain both RGB and depth images were used. Classification accuracies of the point cloud data created from both original and synthetic depth images using various PointNet architectures were measured for sign language recognition. From the original and synthetic point clouds, frame based, Point Gesture Map and Long Short Term Memory data models were used for classification and their performances were compared. In the results, both original and synthetic based data achieved acceptable performance in most models. In general, original depth based point cloud models performed better than synthetic ones, however in some models synthetic depth based models performed better than the originals.

Figures

Figures reproduced from arXiv: 2608.09400 by the authors.

Figure 1
Figure 1. PointNet architecture. [6]. Various networks can be used for creating depth images from monocular RGB images. Among the architectures, there are models based on Stable Diffusion [22], as well as discriminative models like Depth Anything V2 [23]. In Depth Anything V2 model, a DINO-v2 transformer model with a discriminative generative neural network is trained using synthetic images. This network produced depth images… view at source ↗
Figure 2
Figure 2. Example data from the Dataset-A [24]. Dataset-B (KArSL): Dataset-B is a dataset of gestures for the Arabic Sign Language [25]. It was recorded using Microsoft Kinect V2 camera and contains RGB, depth and skeleton data. Dataset was recorded by three signers, performing 502 different gestures with a green screen background. Each gesture was repeated for 50 times. Dataset contains 75300 samples of gestures in video fra… view at source ↗
Figure 3
Figure 3. Example data from the Dataset-B [25] [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (21 more)
Figure 4
Figure 4. Figure 4: Example data from the Dataset-C [26]. 3.2 Data Preparation Dataset-A (Real-time ASL Fingerspelling): Data from Dataset-A is inside the folders named from a to y, represent￾ing each gesture. Inside these folders, RGB and depth images of the each gesture is located. All …
Figure 5
Figure 5. Figure 5: Examples from the data created from Dataset-A. Top two rows: (a, g) Original depth, (b, h) Raw point [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Frame interpolation for the LSTM. GlobalMaxPooling1D layer of the Keras implementation. %75 of the original point cloud based LSTM data (55445 samples) was used for training, %25 remaining samples (18482 samples) were reserved for testing. For the synthetic point cloud…
Figure 7
Figure 7. Figure 7: Examples from the data created from Dataset-B. Top two rows: (a, g) Original depth, (b, h) Raw point [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Examples from the data created from Dataset-C. Top two rows: (a, g) Original depth, (b, h) Raw point [PITH_FULL_IMAGE:figures/full_fig_p008_8.png]
Figure 9
Figure 9. Figure 9: Training for Dataset-A original depth frames. [PITH_FULL_IMAGE:figures/full_fig_p009_9.png]
Figure 11
Figure 11. Figure 11: Training for Dataset-A original PGM depth [PITH_FULL_IMAGE:figures/full_fig_p009_11.png]
Figure 13
Figure 13. Figure 13: Training for Dataset-B original depth frames. [PITH_FULL_IMAGE:figures/full_fig_p010_13.png]
Figure 15
Figure 15. Figure 15: Training for Dataset-B original depth frames [PITH_FULL_IMAGE:figures/full_fig_p010_15.png]
Figure 17
Figure 17. Figure 17: Training for Dataset-B original depth frames [PITH_FULL_IMAGE:figures/full_fig_p010_17.png]
Figure 19
Figure 19. Figure 19: Training for Dataset-C original depth frames. [PITH_FULL_IMAGE:figures/full_fig_p011_19.png]
Figure 21
Figure 21. Figure 21: Training for Dataset-C original depth frames [PITH_FULL_IMAGE:figures/full_fig_p011_21.png]
Figure 23
Figure 23. Figure 23: Training for Dataset-C original depth frames [PITH_FULL_IMAGE:figures/full_fig_p011_23.png]
Figure 25
Figure 25. Figure 25: Confusion matrix for Dataset-A original depth frames. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p012_25.png]
Figure 27
Figure 27. Figure 27: Confusion matrix for Dataset-A original PGM depth frames. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p012_27.png]
Figure 29
Figure 29. Figure 29: Confusion matrix for Dataset-B original depth frames. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p013_29.png]
Figure 31
Figure 31. Figure 31: Confusion matrix for Dataset-B original depth frames PGM. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p013_31.png]
Figure 33
Figure 33. Figure 33: Confusion matrix for Dataset-B original depth frames LSTM. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p013_33.png]
Figure 35
Figure 35. Figure 35: Confusion matrix for Dataset-C original depth frames. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p015_35.png]
Figure 37
Figure 37. Figure 37: Confusion matrix for Dataset-C original depth frames PGM. Predicted Label True Label [PITH_FULL_IMAGE:figures/full_fig_p015_37.png]
Figure 39
Figure 39. Figure 39: Confusion matrix for Dataset-C synthetic depth frames LSTM. [PITH_FULL_IMAGE:figures/full_fig_p015_39.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references · 23 canonical work pages

  1. [18]

    Pseudodepth-slr: Generating depth data for sign language recogni- tion,

    N. Sarhan, J. M. Willruth, and S. Fritnrop, “Pseudodepth-slr: Generating depth data for sign language recogni- tion,” inInternational Conference on Computer Vision Systems. Springer, 2023, pp. 51–62

  2. [1]

    Deep learning-based 3d point cloud classification: A systematic survey and outlook,

    H. Zhang, C. Wang, S. Tian, B. Lu, L. Zhang, X. Ning, and X. Bai, “Deep learning-based 3d point cloud classification: A systematic survey and outlook,”Displays, vol. 79, p. 102456, 2023

  3. [2]

    A review of deep learning-based semantic segmentation for point cloud,

    J. Zhang, X. Zhao, Z. Chen, and Z. Lu, “A review of deep learning-based semantic segmentation for point cloud,” IEEE access, vol. 7, pp. 179 118–179 133, 2019

  4. [3]

    Object volume estimation based on 3d point cloud,

    W.-C. Chang, C.-H. Wu, Y .-H. Tsai, and W.-Y . Chiu, “Object volume estimation based on 3d point cloud,” in 2017 International automatic control conference (CACS). IEEE, 2017, pp. 1–5

  5. [4]

    Space-time event clouds for gesture recognition: From rgb cameras to event cameras,

    Q. Wang, Y . Zhang, J. Yuan, and Y . Lu, “Space-time event clouds for gesture recognition: From rgb cameras to event cameras,” in2019 IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE, 2019, pp. 1826–1835

  6. [5]

    A pointnet-based solution for 3d hand gesture recognition,

    R. Mirsu, G. Simion, C. D. Caleanu, and I. M. Pop-Calimanu, “A pointnet-based solution for 3d hand gesture recognition,”Sensors, vol. 20, no. 11, p. 3226, 2020

  7. [6]

    Pointnet: Deep learning on point sets for 3d classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “Pointnet: Deep learning on point sets for 3d classification and segmentation,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2017, pp. 652–660

  8. [7]

    A review of hand gesture and sign language recognition techniques,

    M. J. Cheok, Z. Omar, and M. H. Jaward, “A review of hand gesture and sign language recognition techniques,” International Journal of Machine Learning and Cybernetics, vol. 10, no. 1, pp. 131–153, 2019

Show all 30 references
  1. [8]

    Quantitative survey of the state of the art in sign language recognition,

    O. Koller, “Quantitative survey of the state of the art in sign language recognition,”arXiv preprint arXiv:2008.09918, 2020

  2. [9]

    Unraveling a decade: A comprehensive survey on isolated sign language recogni- tion,

    N. Sarhan and S. Frintrop, “Unraveling a decade: A comprehensive survey on isolated sign language recogni- tion,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 3210–3219

  3. [10]

    Sign language recognition: A comprehensive review of traditional and deep learning approaches, datasets, and challenges,

    T. Tao, Y . Zhao, T. Liu, and J. Zhu, “Sign language recognition: A comprehensive review of traditional and deep learning approaches, datasets, and challenges,”Ieee Access, vol. 12, pp. 75 034–75 060, 2024

  4. [11]

    Advancements in sign language recognition: A com- prehensive review and future prospects,

    B. A. Al Abdullah, G. A. Amoudi, and H. S. Alghamdi, “Advancements in sign language recognition: A com- prehensive review and future prospects,”IEEE Access, vol. 12, pp. 128 871–128 895, 2024

  5. [12]

    Deep learn- ing approaches for continuous sign language recognition: A comprehensive review,

    A. Khan, S. Jin, G.-H. Lee, G. E. Arzu, L. M. Dang, T. N. Nguyen, W. Choi, and H. Moon, “Deep learn- ing approaches for continuous sign language recognition: A comprehensive review,”IEEE Access, vol. 13, pp. 55 524–55 544, 2025

  6. [13]

    Large-scale isolated gesture recognition using convolutional neural networks,

    P. Wang, W. Li, S. Liu, Z. Gao, C. Tang, and P. Ogunbona, “Large-scale isolated gesture recognition using convolutional neural networks,” in2016 23rd international conference on pattern recognition (ICPR). IEEE, 2016, pp. 7–12

  7. [14]

    Recognition of fingerspelling sequences in polish sign language using point clouds obtained from depth images,

    D. Warchoł, T. Kapu ´sci´nski, and M. Wysocki, “Recognition of fingerspelling sequences in polish sign language using point clouds obtained from depth images,”Sensors, vol. 19, no. 5, p. 1078, 2019

  8. [15]

    User-independent american sign language alphabet recognition based on depth image and pcanet features,

    W. Aly, S. Aly, and S. Almotairi, “User-independent american sign language alphabet recognition based on depth image and pcanet features,”IEEE Access, vol. 7, pp. 123 138–123 150, 2019

  9. [16]

    A comprehensive study on deep learning-based methods for sign language recognition,

    N. Adaloglou, T. Chatzis, I. Papastratis, A. Stergioulas, G. T. Papadopoulos, V . Zacharopoulou, G. J. Xydopoulos, K. Atzakas, D. Papazachariou, and P. Daras, “A comprehensive study on deep learning-based methods for sign language recognition,”IEEE transactions on multimedia, ...

  10. [17]

    Isolated sign language recognition with depth cameras,

    M. Oszust and J. Krupski, “Isolated sign language recognition with depth cameras,”Procedia Computer Science, vol. 192, pp. 2085–2094, 2021

  11. [19]

    Open3D: A modern library for 3D data processing,

    Q.-Y . Zhou, J. Park, and V . Koltun, “Open3D: A modern library for 3D data processing,”arXiv:1801.09847, 2018

  12. [20]

    Evaluation of hand washing procedure using vision-based frame level and spatio- temporal level data models,

    R. Özakar and E. Gedikli, “Evaluation of hand washing procedure using vision-based frame level and spatio- temporal level data models,”Electronics, vol. 12, no. 9, p. 2024, 2023

  13. [21]

    Long short-term memory,

    S. Hochreiter and J. Schmidhuber, “Long short-term memory,”Neural computation, vol. 9, no. 8, pp. 1735–1780, 1997

  14. [22]

    High-resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High-resolution image synthesis with latent diffusion models,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 684–10 695. 16 Sign Language Recognition Using Origina...

  15. [23]

    Depth anything v2,

    L. Yang, B. Kang, Z. Huang, Z. Zhao, X. Xu, J. Feng, and H. Zhao, “Depth anything v2,”Advances in Neural Information Processing Systems, vol. 37, pp. 21 875–21 911, 2024

  16. [24]

    Spelling it out: Real-time asl fingerspelling recognition,

    N. Pugeault and R. Bowden, “Spelling it out: Real-time asl fingerspelling recognition,” in2011 IEEE Interna- tional conference on computer vision workshops (ICCV workshops). Ieee, 2011, pp. 1114–1119

  17. [25]

    Karsl: Arabic sign language database,

    A. A. I. Sidig, H. Luqman, S. Mahmoud, and M. Mohandes, “Karsl: Arabic sign language database,”ACM Transactions on Asian and Low-Resource Language Information Processing (TALLIP), vol. 20, no. 1, pp. 1–19, 2021

  18. [26]

    Autsl: A large scale multi-modal turkish sign language dataset and baseline methods,

    O. M. Sincan and H. Y . Keles, “Autsl: A large scale multi-modal turkish sign language dataset and baseline methods,”IEEE access, vol. 8, pp. 181 340–181 355, 2020

  19. [27]

    (2026) Tensorflow-keras pointnet

    Tensorflow-Keras. (2026) Tensorflow-keras pointnet. Accessed: 2026-06-19. [Online]. Available: https: //keras.io/examples/vision/pointnet/

  20. [28]

    Adam: A method for stochastic optimization,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,”arXiv preprint arXiv:1412.6980, 2014

  21. [29]

    R. Ozakar. (2026) Github repo. Accessed: 2026-08-10. [Online]. Available: https://github.com/r-ozakar

  22. [30]

    Hand-model-aware sign language recognition,

    H. Hu, W. Zhou, and H. Li, “Hand-model-aware sign language recognition,” inProceedings of the AAAI confer- ence on artificial intelligence, vol. 35, no. 2, 2021, pp. 1558–1566. 17

Pith tools

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