Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Image2CADSeq: Computer-Aided Design Sequence and Knowledge Inference from Product Images

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

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

Pith's one-line read A single 2D image can be converted into an editable CAD construction sequence, and this paper presents the first model—Image2CADSeq—that attempts it with a target-embedding variational autoencoder.

desk verdict A credible proof of concept for image-to-CAD-sequence that stays entirely inside its five training templates; the external validity claims need more work. read the letter →

arxiv 2501.04928 v1 pith:ET3B4VH7 submitted 2025-01-09 cs.CV cs.AI

Xingang Li , Zhenghui Sha This is my paper · ORCID
classification cs.CVcs.AI
keywords Image2CADSeqCADsequencepredictionreverseengineeringtarget-embeddingvariationalautoencoderSketch-and-Extrudedesignrulelearningimage-to-CADFusion360Gallery
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper attempts the first direct prediction of a CAD construction sequence—the ordered Sketch-and-Extrude operations behind a 3D model—from a single 2D image. The authors build the Image2CADSeq model, which encodes a CAD program as a fixed-length matrix of quantized operation tokens, learns a latent space for those programs with a variational autoencoder, then trains a ResNet image encoder to regress that latent space. On a synthesized dataset of five template shapes, the target-embedding VAE variant outperforms the autoencoder baseline, especially when the training data embeds design rules that link parameters. The authors also build a multi-level evaluation framework for CAD sequence prediction. If the approach generalizes, it would let designers and even non-experts reconstruct editable, parametric CAD models directly from photographs.

What carries the argument

The load-bearing machinery is a vectorized design representation plus a two-stage target-embedding architecture. Each CAD program is simplified into a Sim-Gallery DSL, then written as a 10x7 feature matrix whose rows are CAD operation vectors [t, I, x, y, α, r, d], with continuous parameters quantized into 256 bins and unused slots filled with −1; start and end markers pad every program to fixed length. In Stage 1, a transformer-based variational autoencoder learns a 256-dimensional latent space for these matrices by reconstruction with KL divergence. In Stage 2, a pretrained ResNet18 with dropout regresses that latent space from the input image, so that at inference the Stage 2 encoder plus the Stage 1 decoder outputs a feature matrix that a parser converts into a Gallery DSL program and then into a solid model.

What would settle it

Take a photograph of an object outside the five template classes—or of a template viewed from a different angle—and run it through the trained Image2CADSeq model. If parsing rate and shape-category accuracy fall to near random levels while synthetic-template performance stays high, the central claim of generalization from images to CAD sequences is falsified for any input distribution beyond the training setup.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that CAD sequences can be reverse-engineered from ordinary 2D images, not just from point clouds, voxels, or B-rep models. The discovery is that a target-embedding variational autoencoder (TEVAE), pretrained to reconstruct vectorized CAD programs in a latent space and then coupled to an image encoder, can predict Sketch-and-Extrude sequences whose rendered 3D models match the input shape: with design-rule-embedded training data, the TEVAE reaches the best ACP, ASOT, AOT, AP1, and MSOT scores of the three cases tested, and its parsed models achieve higher IoU and lower MSE than the TEA baseline. The paper also demonstrates that embedding design rules in the synthetic data substantially improves prediction, and it reports a real-world validation in which smartphone photos of 3D-printed templates parse at a 70% rate, with four of ten objects assigned the correct shape category though parameters remain inaccurate.

Load-bearing premise

The approach assumes that training on renders of five simple template shapes, all viewed from the same camera position, teaches enough about images and CAD programs to handle arbitrary product photos; the paper's own real-world test shows the assumption is only partly holding.

Editorial extensions

If this is right

  • If the central claim holds, reverse engineering can start from a photograph rather than a 3D scan, removing a major data-acquisition bottleneck.
  • CAD sequences, unlike B-rep meshes, preserve the modeling history, so reconstructed models become editable and parametric rather than static geometry.
  • Embedding design rules into synthetic training data measurably improves operation-type and parameter prediction, suggesting rule-aware data synthesis is a viable route to better models.
  • The multi-level evaluation framework (ACP, ASOT/AOT, AP1, MSOT, IoU, MSE) gives later image-to-CAD-sequence work a common yardstick.
  • A working single-image-to-CAD pipeline would let non-experts capture a physical part and obtain a starting CAD model for modification, supporting design democratization.

Reading between the lines

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

  • The fixed camera setup (20,20,20) and five template shapes mean the current results likely overstate performance on arbitrary viewpoints, materials, and geometries; a viewpoint-perturbed test set would quantify this gap.
  • The 70% parsing rate on real photos suggests the main bottleneck is the image-to-latent alignment in Stage 2, not CAD-sequence reconstruction, so cross-modal alignment techniques are the natural next lever.
  • Extending the DSL with Revolve, Sweep, and Fillet, and adding assemblies, could turn the same pipeline from primitive shapes into industrially useful reconstruction.
  • One testable extension: if rule-embedded synthesis is what drives the gain, then sampling parameters from learned design constraints rather than hand-coded rules should push parameter accuracy further.
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 introduces Image2CADSeq, a two-stage neural architecture that predicts a CAD program (a sequence of Sketch-and-Extrude operations in a simplified DSL) from a single input image. The authors synthesize a dataset of 22,000 rendered images from five template shapes with and without embedded "design rules," compare a target-embedding autoencoder (TEA) with a target-embedding variational autoencoder (TEVAE), and propose a multi-level evaluation framework (ACP, ASOT, AOT, AP1, MSOT, plus IoU/MSE for geometry). They report that TEVAE trained on the rule-based dataset outperforms the TEA baselines on their metrics, and they validate the model on smartphone photographs of 3D-printed copies of the same five template shapes, obtaining a 70% parsing rate.

Significance. If the central claims held, the work would be a useful proof-of-concept for a novel task: recovering editable CAD construction histories from images rather than from 3D data. The paper has concrete strengths: it ships a complete data-synthesis pipeline, a simplified DSL with a reversible parsing path to Fusion 360 Gallery, a transparent comparison of two architectures, and a custom multi-level evaluation framework that includes a random-guess parameter baseline (Section 5.3.2, Eq. (8)). The authors also state limitations honestly in Section 6.3. However, the significance as an advance toward the stated goal of reverse-engineering CAD sequences from arbitrary product images is limited by the narrow, in-distribution evaluation: all quantitative experiments stay within the five template shape categories and a single camera viewpoint, so the results do not support the broad claims in the abstract and introduction.

major comments (4)
  1. [§5.1, §5.3, §6.3] The evaluation never leaves the five training templates. All 22,000 synthetic shapes are generated from the five template programs in Table 5 and rendered from the single fixed camera position (20,20,20); the train/validation/test split is random, not categorical. The real-world validation in Section 6.3 uses 3D-printed copies of the same five template categories. Consequently, no experiment requires the model to predict a CAD sequence for an unseen topology or an unseen viewpoint, and correct operation-type sequences can be obtained by solving a 5-way classification problem. The abstract's claim that the model can generate CAD sequences from 2D image data of products is therefore unsupported for arbitrary images; the paper should either add held-out shape categories/viewpoints/lighting conditions to the evaluation or substantially narrow the stated claims to in-distribution proof-of-concept.
  2. [§4.4, §5.1, §5.3.3] The "with-rules" dataset embeds hand-defined correlations between parameters (e.g., extrusion depth of a circle determined by center coordinates), and Section 6.1 credits the model with learning these rules. This is partially circular: the model is trained and tested on data generated from the same hand-written rules, so high AP1 values reflect the model exploiting correlations the authors themselves introduced. To support the claim that the model learns transferable design knowledge, the authors should evaluate on rule-consistent versus rule-agnostic test sets, or ablate the rules from the test distribution and show performance rather than regressing to the random baseline.
  3. [§5.3, Table 6, Fig. 5] All reported results are point estimates from a single random split with no error bars, confidence intervals, or repeated seeds. Given the stochastic training procedure (dropout, VAE sampling, random initialization), the claimed superiority of TEVAE over TEA could be within run-to-run variation. The paper should report results over multiple seeds (at least 3-5) and, ideally, a paired statistical test across seeds to support the architecture comparison.
  4. [§6.3, Fig. 10] The real-world validation reports a parsing rate of 70%, with only four of ten objects assigned the correct shape category and the authors noting inaccurate parameters. This is a large degradation from the synthetic results and directly affects the headline claim of "great potential ... from product images." The paper should provide a more detailed analysis of the failure cases (e.g., which categories confound, whether failures are in operation-type prediction or parameter regression, how viewpoint/background/lighting affect performance) and should temper the concluding claims accordingly.
minor comments (5)
  1. [Eq. (1)] The equation uses "bPi" where the text refers to the predicted program \hat{P}_i; this is likely a LaTeX rendering issue and should be fixed.
  2. [§4.2] Setting the profile index [I], Boolean operation O, and scale factor s to fixed defaults (0, 3, 10) silently discards information that is relevant to real CAD programs. The paper should explicitly justify why these parameters are ignorable for the evaluated shapes, particularly the Boolean operation, which determines whether extrusion adds or cuts material.
  3. [§5.3.2, Eq. (9)] The derivation of the random baseline that includes the Sketch parameter introduces a fraction 11/91 and a coefficient 80/91, but the text does not explain how these numbers are obtained from the dataset composition. Please provide the counting argument.
  4. [§4.5, Table 4] AP2 is defined in the evaluation framework but never reported or discussed in the experimental results. If it is omitted for a stated reason (e.g., matching ambiguity with repeated operation types), say so explicitly; otherwise, report it.
  5. [§5.2] The hyperparameter search is described only for Stage 1; the Stage 2 choices (learning rate 0.0001, dropout 0.4, 50 epochs) are given without sensitivity analysis, so it is unclear how robust the reported results are to these choices.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all load-bearing results are empirical evaluations rather than derivations from inputs.

full rationale

The Image2CADSeq paper makes no analytic claim whose output equals its input. The central results are empirical measurements on held-out splits of a synthetic dataset, and the paper's own real-world validation (Section 6.3) reports degraded performance, which is inconsistent with a prediction that is forced by construction. The TEVAE architecture is taken from the authors' prior work [10], but its effectiveness is evaluated in this paper against a TEA baseline rather than assumed from the citation, so the self-citation is not load-bearing. The 'design rules' embedded in the synthetic dataset are an input data-construction choice, not a parameter fitted to the evaluation metrics; the test-set predictions are not statistically forced by the rules themselves. The evaluation metrics are self-defined, but they are used to compare predicted outputs with ground-truth programs, not as derivation inputs. No equation in the paper reduces to its own input, and no fitted parameter is renamed as a prediction. The acknowledged limitation that only 70% of real photos parse and only 4 of 10 are assigned the correct shape category further confirms that the model does not trivially succeed outside the synthetic training distribution. Therefore no circularity is present.

Assumptions & free parameters 9 free parameters · 6 assumptions · 2 invented entities

The central claim rests largely on the authors' own synthesized data, hand-chosen representation, and custom metrics. The two most consequential free choices are the 5-template training distribution with a fixed camera and the design rules embedded in the data; together they make the reported operation-type accuracy (96.7 percent ASOT) a within-distribution pattern-matching result rather than evidence of general design-knowledge inference.

free parameters (9)
  • latent dimension size = 256
    Selected by grid search over 64, 128, 256, 512 as the value with lowest reconstruction loss on the test set (Section 5.2).
  • tolerance eta for ACP and AP1 = 3
    Chosen by hand (Section 5.3) to allow deviations among 256 quantization levels; results depend strongly on this value.
  • quantization levels per continuous parameter = 256
    Hand-chosen in Section 4.2 to represent continuous parameters as 8-bit integers; coarser or finer quantization changes task difficulty and metrics.
  • maximum program length Nc = 10
    Set to 10 in Section 4.2; truncates or pads all CAD programs, limiting complexity to at most 10 operations.
  • default profile index, boolean operation, scale factor = [I]=0, O=3, s=10
    Set as constants in Section 4.2; models do not predict profile indexing, Boolean operations, or model scale, simplifying the task.
  • template shapes and parameter ranges = 5 template shapes; ranges in Table 3
    The training distribution is entirely generated from these templates and ranges (Section 5.1); the model cannot represent other construction sequences.
  • camera pose for rendering = (20, 20, 20) looking at origin
    All training images use this uniform perspective camera (Section 5.1); real photos from other viewpoints are out-of-distribution.
  • design rules in the with-rules dataset = e.g., extrusion depth of a circle determined by its center point
    Hand-embedded rules create the latent patterns that the model is later credited with learning (Sections 5.1 and 6.1).
  • training hyperparameters = 500 epochs Stage 1, 50 epochs Stage 2, batch 512/128, lr 0.001/0.0001, dropout 0.4
    Chosen by grid search and heuristic in Section 5.2; results are tied to these settings.
assumptions (6)
  • standard math Transformer-based VAE/AE and ResNet18 architectures behave as described in prior literature (DeepCAD, TEVAE, ResNet).
    The method builds on these published models without re-deriving them (Sections 3 and 5.2).
  • domain assumption A single fixed-perspective render is sufficient to determine the CAD parameters (sketch plane, dimensions, extrusion depth).
    All synthetic training uses one camera view (Section 5.1); if multiple views or depth cues are needed, the task is underdetermined.
  • domain assumption The five template shapes and the Gallery DSL's Sketch-and-Extrude grammar cover the intended product-image application space.
    Data synthesis uses 5 templates and only Sketch and Extrude operations (Section 4.1 and Table 5); real-world assemblies are acknowledged as outside scope in Section 6.3.
  • ad hoc to paper The custom evaluation metrics (ACP, ASOT, AOT, AP1, MSOT) are valid measures of CAD sequence quality.
    These metrics are introduced in Section 4.5 without external validation against human judgments or engineering usefulness.
  • ad hoc to paper The hand-defined design rules embedded in the with-rules dataset are representative of real-world design knowledge.
    Section 5.1 defines rules such as extrusion depth depending on circle center coordinates; the improvement in Section 6.1 is attributed to these rules.
  • ad hoc to paper Ignoring profile index, Boolean operation, and scale (setting defaults 0, 3, 10) loses no information needed for the evaluated shapes.
    Section 4.2 fixes these variables; this is valid for single-profile additive extrusions but fails for multi-profile or Boolean-cut designs.
invented entities (2)
  • Sim-Gallery DSL
    purpose: A simplified CAD program language used as the neural network target and parser interface.
    Introduced in Section 4.1 by stripping the Fusion 360 Gallery DSL to pure operation functions; it has no external benchmark validating it.
  • Feature matrix P (10x7)
    purpose: Fixed-length vectorized representation of a CAD program used as the VAE input and output.
    Defined in Section 4.2 with quantization and padding; internal to the paper's pipeline.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Image2CADSeq: Computer-Aided Design Sequence and Knowledge Inference from Product Images." pith.science (2026). https://pith.science/paper/ET3B4VH7

@misc{pith2026250104928,
  author       = {Pith},
  title        = {Pith review of: Image2CADSeq: Computer-Aided Design Sequence and Knowledge Inference from Product Images},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ET3B4VH7}},
  note         = {Machine review of arXiv:2501.04928}
}
read the original abstract

Computer-aided design (CAD) tools empower designers to design and modify 3D models through a series of CAD operations, commonly referred to as a CAD sequence. In scenarios where digital CAD files are not accessible, reverse engineering (RE) has been used to reconstruct 3D CAD models. Recent advances have seen the rise of data-driven approaches for RE, with a primary focus on converting 3D data, such as point clouds, into 3D models in boundary representation (B-rep) format. However, obtaining 3D data poses significant challenges, and B-rep models do not reveal knowledge about the 3D modeling process of designs. To this end, our research introduces a novel data-driven approach with an Image2CADSeq neural network model. This model aims to reverse engineer CAD models by processing images as input and generating CAD sequences. These sequences can then be translated into B-rep models using a solid modeling kernel. Unlike B-rep models, CAD sequences offer enhanced flexibility to modify individual steps of model creation, providing a deeper understanding of the construction process of CAD models. To quantitatively and rigorously evaluate the predictive performance of the Image2CADSeq model, we have developed a multi-level evaluation framework for model assessment. The model was trained on a specially synthesized dataset, and various network architectures were explored to optimize the performance. The experimental and validation results show great potential for the model in generating CAD sequences from 2D image data.

Figures

Figures reproduced from arXiv: 2501.04928 by the authors.

Figure 1
Figure 1. Approach overview. 4 METHODOLOGY The flowchart depicted in [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Image2CADSeq model using a target-embedding represen [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Synthesis pipeline for the training dataset of data pairs of image and vectorized CAD sequence, exemplified using a cylinder model [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Implementation of the Image2CADSeq model. Two different [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]
Figure 5
Figure 5. Figure 5: Evaluation of the Image2CADSeq model’s performance using two distinct architectures with two datasets. (a) Case 1: Results from [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: Overall parameter accuracy versus the tolerance levels eval [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: The variation of AP1 versus tolerance for the operation parameters for the [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Parsing rate, intersection over union (IoU), and mean [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Qualitative analysis of Image2CADSeq model using TEVAE. (a) Near-perfect predictions: High accuracy in shape, size, and position. [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Validation experiments using real-world images. (a) 3D-printed design objects; (b) Photographs (3024 [PITH_FULL_IMAGE:figures/full_fig_p018_10.png]

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. Drawing2CAD: Sequence-to-Sequence Learning for CAD Generation from Vector Drawings

    cs.CV 2025-08 conditional novelty 6.0 of 10

    Drawing2CAD generates parametric CAD operation sequences from SVG engineering drawings, outperforming a raster-input baseline and a DeepCAD-vector baseline on accuracy and validity.

Reference graph

Works this paper leans on

53 extracted references · 52 canonical work pages · cited by 1 Pith paper

  1. [1]

    5 - computer- aided design

    Rosato, D., and Rosato, D., 2003. “5 - computer- aided design”. In Plastics Engineered Product Design, D. Rosato and D. Rosato, eds. Elsevier Science, Ams- terdam, pp. 344–380

    work page 2003

  2. [2]

    Deepcad: A deep generative network for computer-aided design models

    Wu, R., Xiao, C., and Zheng, C., 2021. “Deepcad: A deep generative network for computer-aided design models”. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision, pp. 6772–6782

    work page 2021

  3. [3]

    Sketchgen: Gen- erating constrained cad sketches

    Para, W., Bhat, S., Guerrero, P., Kelly, T., Mitra, N., Guibas, L. J., and Wonka, P., 2021. “Sketchgen: Gen- erating constrained cad sketches”. Advances in Neural Information Processing Systems, 34

    work page 2021

  4. [4]

    Re- verse engineering of geometric models—an introduc- tion

    Varady, T., Martin, R. R., and Cox, J., 1997. “Re- verse engineering of geometric models—an introduc- tion”. Computer-aided design, 29(4), pp. 255–268

    work page 1997

  5. [5]

    Reverse engineering modeling methods and tools: a survey

    Buonamici, F., Carfagni, M., Furferi, R., Governi, L., Lapini, A., and V olpe, Y ., 2018. “Reverse engineering modeling methods and tools: a survey”. Computer- Aided Design and Applications, 15(3), pp. 443–464

    work page 2018

  6. [6]

    Point2cyl: Reverse engineering 3d objects from point clouds to extrusion cylinders

    Uy, M. A., Chang, Y .-Y ., Sung, M., Goel, P., Lam- bourne, J. G., Birdal, T., and Guibas, L. J., 2022. “Point2cyl: Reverse engineering 3d objects from point clouds to extrusion cylinders”. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp. 11850–11860

    work page 2022

  7. [7]

    Extrudenet: Unsupervised inverse sketch-and-extrude for shape parsing

    Ren, D., Zheng, J., Cai, J., Li, J., and Zhang, J., 2022. “Extrudenet: Unsupervised inverse sketch-and-extrude for shape parsing”. In European Conference on Com- puter Vision, Springer, pp. 482–498

    work page 2022

  8. [8]

    Re- constructing editable prismatic cad from rounded voxel models

    Lambourne, J. G., Willis, K., Jayaraman, P. K., Zhang, L., Sanghi, A., and Malekshan, K. R., 2022. “Re- constructing editable prismatic cad from rounded voxel models”. In SIGGRAPH Asia 2022 Conference Papers, pp. 1–9

    work page 2022

Show all 53 references
  1. [9]

    Secad-net: Self-supervised cad reconstruction by learning sketch-extrude operations

    Li, P., Guo, J., Zhang, X., and Yan, D.-M., 2023. “Secad-net: Self-supervised cad reconstruction by learning sketch-extrude operations”. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 16816–16826

    2023

  2. [10]

    A predictive and generative design approach for three-dimensional mesh shapes using target-embedding variational au- toencoder

    Li, X., Xie, C., and Sha, Z., 2022. “A predictive and generative design approach for three-dimensional mesh shapes using target-embedding variational au- toencoder”. Journal of Mechanical Design, 144(11), p. 114501

    2022

  3. [11]

    Deep generative models on 3d representa- tions: A survey

    Shi, Z., Peng, S., Xu, Y ., Geiger, A., Liao, Y ., and Shen, Y ., 2022. “Deep generative models on 3d representa- tions: A survey”. arXiv preprint arXiv:2210.15663

    2022 arXiv

  4. [12]

    Deep learning methods of cross-modal tasks for conceptual design of product shapes: A review

    Li, X., Wang, Y ., and Sha, Z., 2023. “Deep learning methods of cross-modal tasks for conceptual design of product shapes: A review”. Journal of Mechanical De- sign, 145(4), p. 041401

    2023

  5. [13]

    Csg-stump: A learning friendly csg-like representation for interpretable shape parsing

    Ren, D., Zheng, J., Cai, J., Li, J., Jiang, H., Cai, Z., Zhang, J., Pan, L., Zhang, M., Zhao, H., and Yi, S., 2021. “Csg-stump: A learning friendly csg-like representation for interpretable shape parsing”. 2021 IEEE/CVF International Conference on Computer Vi- sion (ICCV), nul...

    2021

  6. [14]

    Csgnet: Neural shape parser for constructive solid geometry

    Sharma, G., Goyal, R., Liu, D., Kalogerakis, E., and Maji, S., 2017. “Csgnet: Neural shape parser for constructive solid geometry”. 2018 IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, null, pp. 5515–5523

    2017

  7. [15]

    Neural shape parsers for constructive solid geometry

    Sharma, G., Goyal, R., Liu, D., Kalogerakis, E., and Maji, S., 2019. “Neural shape parsers for constructive solid geometry”. IEEE Transactions on Pattern Analy- sis and Machine Intelligence, 44, pp. 2628–2640

    2019

  8. [16]

    Ucsg-net-unsupervised discovering of constructive solid geometry tree

    Kania, K., Zieba, M., and Kajdanowicz, T., 2020. “Ucsg-net-unsupervised discovering of constructive solid geometry tree”. Advances in Neural Information Processing Systems, 33, pp. 8776–8786

    2020

  9. [17]

    Engineering sketch genera- tion for computer-aided design

    Willis, K. D., Jayaraman, P. K., Lambourne, J. G., Chu, H., and Pu, Y ., 2021. “Engineering sketch genera- tion for computer-aided design”. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, pp. 2105–2114

    2021

  10. [18]

    Seff, A., Zhou, W., Richardson, N., and Adams, R. P.,

  11. [19]

    Computer-aided design as language

    Ganin, Y ., Bartunov, S., Li, Y ., Keller, E., and Saliceti, S., 2021. “Computer-aided design as language”. Ad- vances in Neural Information Processing Systems, 34

    2021

  12. [20]

    Discovering design con- cepts for cad sketches

    Yang, Y ., and Pan, H., 2022. “Discovering design con- cepts for cad sketches”. ArXiv, abs/2210.14451, p. null

    2022 arXiv

  13. [21]

    Pie- net: Parametric inference of point cloud edges

    Wang, X., Xu, Y ., Xu, K., Tagliasacchi, A., Zhou, B., Mahdavi-Amiri, A., and Zhang, H., 2020. “Pie- net: Parametric inference of point cloud edges”. Ad- vances in neural information processing systems, 33, pp. 20167–20178

    2020

  14. [22]

    Parsenet: A parametric surface fitting network for 3d point clouds

    Sharma, G., Liu, D., Maji, S., Kalogerakis, E., Chaud- huri, S., and M ˇech, R., 2020. “Parsenet: A parametric surface fitting network for 3d point clouds”. In Com- puter Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part VII 16, Spr...

    2020

  15. [23]

    Smirnov, D., Bessmeltsev, M., and Solomon, J.,

  16. [24]

    Complexgen: Cad reconstruction by b- rep chain complex generation

    Guo, H., Liu, S., Pan, H., Liu, Y ., Tong, X., and Guo, B., 2022. “Complexgen: Cad reconstruction by b- rep chain complex generation”. ACM Transactions on Graphics (TOG), 41(4), pp. 1–18

    2022

  17. [25]

    Solidgen: An autoregressive model for direct b-rep synthesis

    Jayaraman, P. K., Lambourne, J. G., Desai, N., Willis, K., Sanghi, A., and Morris, N. J., 2022. “Solidgen: An autoregressive model for direct b-rep synthesis”.Trans- actions on Machine Learning Research

    2022

  18. [26]

    Neural face identification in a 2d wireframe projection of a man- ifold object

    Wang, K., Zheng, J., and Zhou, Z., 2022. “Neural face identification in a 2d wireframe projection of a man- ifold object”. In Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pp. 1622–1631

    2022

  19. [27]

    Skex- gen: Autoregressive generation of cad construction se- quences with disentangled codebooks

    Xu, X., Willis, K. D., Lambourne, J. G., Cheng, C.- Y ., Jayaraman, P. K., and Furukawa, Y ., 2022. “Skex- gen: Autoregressive generation of cad construction se- quences with disentangled codebooks”. In International Conference on Machine Learning, PMLR, pp. 24698– 24724

    2022

  20. [28]

    Attention is all you need

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

    2017

  21. [29]

    D., Pu, Y ., Luo, J., Chu, H., Du, T., Lam- bourne, J

    Willis, K. D., Pu, Y ., Luo, J., Chu, H., Du, T., Lam- bourne, J. G., Solar-Lezama, A., and Matusik, W.,

  22. [30]

    Inferring cad modeling sequences using zone graphs

    Xu, X., Peng, W., Cheng, C.-Y ., Willis, K. D., and Ritchie, D., 2021. “Inferring cad modeling sequences using zone graphs”. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 6062–6070

    2021

  23. [31]

    Sketch2cad: Sequential cad modeling by sketching in context

    Li, C., Pan, H., Bousseau, A., and Mitra, N. J., 2020. “Sketch2cad: Sequential cad modeling by sketching in context”. ACM Transactions on Graphics (TOG), 39(6), pp. 1–14

    2020

  24. [32]

    Fusion 360 gallery: A dataset and environment for programmatic cad construction from human design sequences

    “Fusion 360 gallery: A dataset and environment for programmatic cad construction from human design sequences”. ACM Transactions on Graphics (TOG), 40(4), pp. 1–24

  25. [33]

    Reduc- ing the dimensionality of data with neural networks

    Hinton, G. E., and Salakhutdinov, R. R., 2006. “Reduc- ing the dimensionality of data with neural networks”. science, 313(5786), pp. 504–507

    2006

  26. [34]

    Auto-encoding variational bayes

    Kingma, D. P., and Welling, M., 2014. “Auto-encoding variational bayes.”. In International Conference on Learning Representations, Y . Bengio and Y . LeCun, eds

    2014

  27. [35]

    Free2cad: Parsing freehand drawings into cad com- mands

    Li, C., Pan, H., Bousseau, A., and Mitra, N. J., 2022. “Free2cad: Parsing freehand drawings into cad com- mands”. ACM Transactions on Graphics (TOG), 41(4), pp. 1–16

    2022

  28. [36]

    Representation learning techniques: An overview

    Khastavaneh, H., and Ebrahimpour-Komleh, H., 2019. “Representation learning techniques: An overview”. In The 7th International Conference on Contemporary Is- sues in Data Science, Springer, pp. 89–104

    2019

  29. [37]

    Variational autoencoders learn trans- ferrable representations of metabolomics data

    Gomari, D. P., Schweickart, A., Cerchietti, L., Paietta, E., Fernandez, H., Al-Amin, H., Suhre, K., and Krum- siek, J., 2022. “Variational autoencoders learn trans- ferrable representations of metabolomics data”. Com- munications Biology, 5(1), p. 645

    2022

  30. [38]

    Rep- resentation learning: A review and new perspectives

    Bengio, Y ., Courville, A., and Vincent, P., 2013. “Rep- resentation learning: A review and new perspectives”. IEEE transactions on pattern analysis and machine in- telligence, 35(8), pp. 1798–1828

    2013

  31. [39]

    Super- vised autoencoders: Improving generalization perfor- mance with unsupervised regularizers

    Le, L., Patterson, A., and White, M., 2018. “Super- vised autoencoders: Improving generalization perfor- mance with unsupervised regularizers”. Advances in neural information processing systems, 31

    2018

  32. [40]

    Learning a predictable and generative vector representation for objects

    Girdhar, R., Fouhey, D. F., Rodriguez, M., and Gupta, A., 2016. “Learning a predictable and generative vector representation for objects”. In European Conference on Computer Vision, Springer, pp. 484–499

    2016

  33. [41]

    Target- embedding autoencoders for supervised representation learning

    Jarrett, D., and van der Schaar, M., 2020. “Target- embedding autoencoders for supervised representation learning”. In International Conference on Learning Representations

    2020

  34. [42]

    Anatomical priors in convolutional networks for unsu- pervised biomedical segmentation

    Dalca, A. V ., Guttag, J., and Sabuncu, M. R., 2018. “Anatomical priors in convolutional networks for unsu- pervised biomedical segmentation”. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 9290–9299

    2018

  35. [43]

    Carlier, A., Danelljan, M., Alahi, A., and Timofte, R.,

  36. [44]

    Mostajabi, M., Maire, M., and Shakhnarovich, G.,

  37. [45]

    Why is tanimoto index an appropriate choice for fingerprint- based similarity calculations?

    Bajusz, D., R ´acz, A., and H ´eberger, K., 2015. “Why is tanimoto index an appropriate choice for fingerprint- based similarity calculations?”. Journal of cheminfor- matics, 7(1), pp. 1–13

    2015

  38. [46]

    Deep residual learning for image recognition

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

    2016

  39. [47]

    Toward hybrid teams: A platform to un- derstand human-computer collaboration during the de- sign of complex engineered systems

    Song, B., Zurita, N. S., Zhang, G., Stump, G., Balon, C., Miller, S., Yukish, M., Cagan, J., and McComb, C., 2020. “Toward hybrid teams: A platform to un- derstand human-computer collaboration during the de- sign of complex engineered systems”. In Proceedings of the Design Soc...

    2020

  40. [48]

    Deepsvg: A hierarchical generative network for vector graphics animation

    “Deepsvg: A hierarchical generative network for vector graphics animation”. Advances in Neural Infor- mation Processing Systems, 33, pp. 16351–16361

  41. [49]

    A guided tour to approximate string matching

    Navarro, G., 2001. “A guided tour to approximate string matching”. ACM computing surveys (CSUR), 33(1), pp. 31–88

    2001

  42. [53]

    Human-centered generative design framework: An early design framework to support concept creation and evaluation

    Demirel, H., Goldstein, M., Li, X., and Sha, Z., 2023. “Human-centered generative design framework: An early design framework to support concept creation and evaluation”. International Journal of Human- Computer Interaction. 20

    2023

  43. [2018]

    Regularizing deep networks by modeling and predicting label structure

    “Regularizing deep networks by modeling and predicting label structure”. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, pp. 5629–5638

  44. [2020]

    Learning manifold patch-based representations of man-made shapes

    “Learning manifold patch-based representations of man-made shapes”. In International Conference on Learning Representations

  45. [2021]

    Vitruvion: A generative model of parametric cad sketches

    “Vitruvion: A generative model of parametric cad sketches”. In International Conference on Learning Representations

Pith tools

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