Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach

T0 review · 2 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 vision-language model rebuilds 3D cabinets from 2D drawings

desk verdict Well-ablated VLM approach to 2D-CAD-to-3D-parametric reconstruction with a genuinely flexible text output, but the raster-vs-vector claim is only validated on simple planks, not the intended cabinet domain. read the letter →

arxiv 2412.11892 v2 pith:RRATFIPC submitted 2024-12-16 cs.CV

Xilin Wang , Jia Zheng , Yuanchao Hu , Hao Zhu , Qian Yu , Zihan Zhou This is my paper · ORCID
classification cs.CV
keywords CADreconstructionvision-languagemodelparametricmodelingrasterimageinputtextshapeprogramorthographicdrawingsannotationlayersautoregressivegeneration
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

CAD2Program claims that 3D parametric models can be reconstructed from 2D CAD drawings by treating the drawing as a raster image and letting a fine-tuned vision-language model write a text program describing the 3D model. The authors show that this pixel-based input matches the accuracy of prior vector-graphics methods on clean three-view drawings, while dropping their requirements: drawings need not be cleaned into a separate geometry layer, and the annotation layer—dimension text and functional symbols—becomes a source of information rather than noise. The output is a Python-style shape program, which the paper argues is more flexible than fixed-size command templates because it naturally handles 373 primitive types and 702 model-specific parameters. On a new dataset of 368K cabinet drawings, the reconstructed 3D boxes reach about 83 percent F1 and model-specific parameters are estimated correctly about 97 percent of the time, with the annotation layer boosting F1 by roughly 20 points. If the claim holds, CAD-to-3D reconstruction becomes less a problem of designing special data representations and more a problem of collecting data and fine-tuning an existing model.

What carries the argument

The central mechanism is a fine-tuned vision-language model with a ViT image encoder, an MLP projector, and a small autoregressive language model, trained to map a rasterized 2D CAD drawing directly to a text shape program. Each primitive instance in that program is written as a `Bbox(...)` line holding the common parameters (position, size, rotation) followed by a model call with its model-specific parameters written as `key=value` pairs, so no fixed-size command template is needed. To help the model identify which primitive a drawing region refers to, each model ID is represented by a special token built from the primitive's name and a rendered snapshot image, encoded with a contrastive vision-language model. The whole pipeline is fine-tuned end-to-end on a dataset of 368K cabinet drawings, with the language model generating the program tokens auto-regressively.

What would settle it

Render a held-out set of cabinets in view layouts and annotation styles never seen in training—e.g., unaligned views, an unusual number of views, or dimension text in a nonstandard font—and compare CAD2Program against a vector-based baseline on those drawings. If the raster-input model's F1 falls far below the vector model's, the claim that pixel input is competitive while imposing fewer restrictions is refuted.

Watch

Extended reading notes

Core claim

The core claim is that a single vision-language model can solve the task end-to-end by treating the input as an image and the output as text. The input side passes the full drawing—both the geometry lines and the annotation symbols—through a ViT image encoder, and the output side auto-regressively generates a text shape program in which each primitive appears as a bounding-box line plus a model call with named parameters, e.g., `model_2 = <model_115813862>(N=1, NKA=928, DBXX=1, BT=18)`. This text representation removes the fixed-size command template used in prior sequence modeling, eliminates parameter quantization by writing values as text, and allows new primitive types to be added without restructuring the output. The paper demonstrates that the vector-specific PlankAssembly system performs comparably when its encoder is swapped for a generic ViT on the same three-view drawings, and that the full CAD2Program trained on cabinet drawings works with a different proxy language and a different vision-language backbone. The result is a method that uses annotations as input rather than stripping them, and that scales to hundreds of primitives and hundreds of model-specific parameters.

Load-bearing premise

The method rests on the premise that a CAD drawing can be read correctly from its pixels alone, so the vision encoder must recover geometry, dimensions, and annotation symbols without any vector entities or layer metadata.

Editorial extensions

If this is right

  • Reconstruction no longer requires a clean geometry layer: drawings can be fed in as-is, and the annotation layer measurably improves accuracy instead of being treated as noise.
  • The text-shaped output removes the fixed-slot constraint, allowing one model to work with 373 primitive types and 702 model-specific parameters without lengthening or restructuring the output sequence.
  • Common parameters are emitted as text rather than discretized tokens, so the quantization error of fixed-bin encoders disappears.
  • Because the output is free-form text, the same approach extends to new product categories and new primitive libraries by collecting a dataset, and the proxy language can be swapped without retraining the core method (Python and YAML give similar results).
  • A general-purpose vision encoder is as effective as a vector-specific encoder for three-view orthographic drawings, as shown when the PlankAssembly encoder is replaced by a ViT.

Reading between the lines

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

  • The same raster-in, text-out recipe could plausibly extend to other engineering documents whose meaning lives in mixed layers—P&ID diagrams, electrical schematics, or dimensioned mechanical drawings—though the paper only demonstrates cabinets.
  • The special token for model IDs uses a rendered snapshot with default parameter values, so primitives whose appearance changes sharply with their parameters could be hard to retrieve; a parameter-aware embedding would be a natural extension.
  • Since the output is executable text, the same fine-tuned weights could be prompted to edit a model program or answer questions about a drawing, effectively turning the model into a CAD assistant—a use the authors flag as future work.
  • The paper's 'fewer restrictions' claim is demonstrated inside the dataset's distribution of layouts and annotations; a stress test with deliberately arbitrary view arrangements and annotation styles would assess how far the generalization extends.
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

2 major / 5 minor

Summary. CAD2Program fine-tunes Mini-InternVL-1.5-2B to reconstruct 3D parametric cabinet models from 2D CAD drawings. The input is the raw rasterized drawing, encoded by a ViT, and the output is an autoregressively generated Python-style text program listing primitive model IDs, common box parameters, and model-specific parameters. The paper introduces a new dataset of 368K cabinet models with 373 primitives and 702 model-specific parameters, and reports experiments on the raster-input assumption (via a ViT variant of PlankAssembly), the usefulness of annotation layers, output representation (command template vs. text), proxy language, and base VLM. The central claims are that a general ViT can match vector-graphics encoders and that a general-purpose text representation handles a large set of primitives more flexibly than fixed-slot command templates.

Significance. The paper's result, if fully established, would be practically significant: it removes the need for vector-format parsing, geometry-layer separation, and exactly three aligned views, and scales reconstruction to hundreds of primitive types. The component-level comparisons are well designed in several respects: the PlankAssembly (ViT) experiment reuses the same code, data, and decoder as the vector baseline; the annotation-layer ablation is performed on the new dataset with matched architecture; and the output-representation comparison holds the architecture fixed while changing only the sequence format. The release of a large cabinet dataset and the use of publicly available base models are additional strengths. The main gap is that the raster-input premise is only tested on the simple PlankAssembly setting, not on the cabinet setting that motivates the paper.

major comments (2)
  1. [§4.2, Figure 6] The only direct evidence for the claim that a raster image encoder is as effective as a vector-graphics encoder comes from PlankAssembly (ViT) on exactly three aligned, annotation-free views on a fixed 512x512 canvas. The cabinet dataset used in Tables 1-4 and Figure 7 is precisely the regime in which the paper claims fewer restrictions (varying view counts, unaligned views, mixed geometry/annotation layers, 373 primitives, reading of dimension text), yet no vector-input baseline is run there. If a vector encoder with layer metadata and coordinate precision outperforms CAD2Program on this dataset, the central claim fails; if it does not, the claim is currently unverified. I request either a vector-based baseline on the cabinet dataset or an oracle-style experiment that feeds exact vector entities and layer metadata into the same decoder, so that the information loss from rasterization is measured in the intended setting.
  2. [Tables 1-4] All quantitative results are reported for a single training run, without standard deviations, confidence intervals, or significance tests. Several comparative statements rest on small margins (e.g., Table 2 retrieval 93.84 vs. 93.36; Table 4 F1 84.90 vs. 82.76, retrieval 93.98 vs. 93.80), and the claim of 'comparable performance' cannot be distinguished from training noise at these margins. Please report at least three seeds and the corresponding variance for the main comparisons, or a paired test on the test set.
minor comments (5)
  1. [Section 1] The text 'Figure 9 illustrates the two layers' should refer to Figure 1, which is the figure that actually shows the geometry and annotation layers; Figure 9 is an appendix figure.
  2. [Listing 1] The parameter 'uCove=18' in Listing 1 is inconsistent with 'uCover' used in Listing 2 and the example in Figure 10; this is likely a typo.
  3. [Section 3.1] The claim that treating values as text avoids quantization error should state the numerical precision of the serialized values in the dataset (e.g., one decimal place), since the text tokenizer itself imposes representational limits; the claim is only true relative to the chosen binning of the command-template baseline.
  4. [§4.2, Figure 6] The traditional-approach curve is taken directly from prior work, but the paper should specify whether the same noise-level definitions and plotting conventions were used for PlankAssembly (ViT) so that the comparison is fair.
  5. [§4.3, Table 2] The interpretation of Table 2 should state explicitly that the command-template comparison covers only the common-parameter subset, since the command-template variant does not predict model-specific parameters; the current presentation implies text output is as effective across the full task.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the raster-input to text-program mapping is validated by holdout tests and a transparent prior-work baseline, not derived from its own outputs.

full rationale

CAD2Program's central derivation is the raster-input-to-text-program mapping, and it is neither defined in terms of its outputs nor fitted to the test labels. The image encoder is a fixed pretrained backbone, and the language model is fine-tuned on a training split and evaluated on a held-out 2K-test split, so the reported retrieval, reconstruction, and parameter accuracies are genuine holdout predictions. The only self-citation is the use of PlankAssembly (Hu et al. 2023) as the vector-input baseline and as the source of the precision/recall/F1 metric. This reliance is transparent ('Results of PlankAssembly (Vector) and the traditional approach are directly taken from (Hu et al. 2023)'), the prior work is externally published, and the raster variant is retrained from its released code and dataset, so the comparison is an independent empirical baseline rather than a result forced by construction. The Section 4.2 PlankAssembly(ViT) comparison uses simplified three-view drawings, and no vector baseline is run on the cabinet dataset; that is an evidentiary gap about external validity, not a circular step. No equation or fitted parameter is reused as a prediction, and no claim is justified solely by a self-citation.

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

The method is an empirical ML system; no physical or mathematical entities are postulated. The special model-ID token is a learned embedding component, not a new scientific entity. The listed free parameters are hand-chosen evaluation or baseline settings that affect the reported numbers, and the listed axioms are the domain assumptions that carry the central claim.

free parameters (3)
  • 3D bounding-box IoU threshold for true positive = 0.5
    Chosen by hand, following PlankAssembly; defines whether a predicted primitive counts as correct and directly controls the reported precision, recall, and F1.
  • Quantization bins for command-template baseline = 1500 bins at 3mm resolution for position/size, 4 bins for angle
    Hand-chosen for the fixed-slot output variant in Table 2; changing these bins changes the baseline's difficulty and therefore affects the claimed parity of the text output.
  • Maximum generated token length = 4096
    Inference limit for CAD2Program; models with more primitives than fit in 4096 tokens would be truncated, so this is a hand-chosen capacity bound on the central result.
assumptions (3)
  • domain assumption A rasterized rendering of a CAD drawing retains the geometric and annotation information needed to recover the parametric 3D model.
    Core premise of the input design, introduced in Section 3.2 where the drawing is treated as an image; tested indirectly in Section 4.2 but not proven for arbitrary drawing formats.
  • domain assumption Autoregressive prediction of text tokens can represent continuous model parameters with acceptable fidelity.
    Core premise of the output design; supported by Tables 2 and 3, but text decoding must still map numbers to floats correctly.
  • domain assumption The evaluation metric, Hungarian-matched 3D bounding-box IoU with threshold 0.5, is a valid measure of reconstruction quality.
    Adopted from PlankAssembly; no error bars or alternative metrics such as volumetric or manufacturing-quality checks are reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach." pith.science (2026). https://pith.science/paper/RRATFIPC

@misc{pith2026241211892,
  author       = {Pith},
  title        = {Pith review of: From 2D CAD Drawings to 3D Parametric Models: A Vision-Language Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RRATFIPC}},
  note         = {Machine review of arXiv:2412.11892}
}
read the original abstract

In this paper, we present CAD2Program, a new method for reconstructing 3D parametric models from 2D CAD drawings. Our proposed method is inspired by recent successes in vision-language models (VLMs), and departs from traditional methods which rely on task-specific data representations and/or algorithms. Specifically, on the input side, we simply treat the 2D CAD drawing as a raster image, regardless of its original format, and encode the image with a standard ViT model. We show that such an encoding scheme achieves competitive performance against existing methods that operate on vector-graphics inputs, while imposing substantially fewer restrictions on the 2D drawings. On the output side, our method auto-regressively predicts a general-purpose language describing 3D parametric models in text form. Compared to other sequence modeling methods for CAD which use domain-specific sequence representations with fixed-size slots, our text-based representation is more flexible, and can be easily extended to arbitrary geometric entities and semantic or functional properties. Experimental results on a large-scale dataset of cabinet models demonstrate the effectiveness of our method.

Figures

Figures reproduced from arXiv: 2412.11892 by the authors.

Figure 1
Figure 1. Illustration of the geometry and annotation layers [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Problem statement. Given (a) a 2D CAD drawing of a product (e.g., a cabinet), our goal is to reconstruct (b) the 3D model of a product. (c) In CAD software, the 3D model is conventionally built by assembling pre-defined primitive models, where (d) each primitive model is defined by a computer program describing its model ID and a number of parameters. that the shelf is adjustable, whereas the red triangle indicates … view at source ↗
Figure 3
Figure 3. Illustration of the model-specific parameters of a [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figures from the paper (6 more)
Figure 5
Figure 5. Figure 5: Dataset statistics. Left: the number of cabinets w.r.t. the number of primitives per cabinet. Right: the num￾ber of primitives w.r.t. the number of model-specific param￾eters per primitive. encoder (Dosovitskiy et al. 2020) and a RoBERTa-wwm￾large text encoder (Cui et …
Figure 7
Figure 7. Figure 7: Qualitative results. For each case, we show [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Examples of the generated engineering drawings. [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: An example CAD drawing. Left: geometry layer only. Right: both the geometry and annotation layers . A.2 Experiments on the Output In this experiment, we train two variants of CAD2PROGRAM. For the first variant which outputs domain-specific se￾quence representations, we…
Figure 10
Figure 10. Figure 10: A example conversation using Python [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: A example conversation using YAML [PITH_FULL_IMAGE:figures/full_fig_p012_11.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

51 extracted references · 43 canonical work pages · cited by 1 Pith paper

  1. [1]

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

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

  2. [2]

    write newline

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

  3. [3]

    Chen, Z.; Wang, W.; Tian, H.; Ye, S.; Gao, Z.; Cui, E.; Tong, W.; Hu, K.; Luo, J.; Ma, Z.; Ma, J.; Wang, J.; Dong, X.; Yan, H.; Guo, H.; He, C.; Shi, B.; Jin, Z.; Xu, C.; Wang, B.; Wei, X.; Li, W.; Zhang, W.; Zhang, B.; Cai, P.; Wen, L.; Yan, X.; Dou, M.; Lu, L.; Zhu, X.; Lu, T.; Lin, D.; Qiao, Y.; Dai, J.; and Wang, W. 2024 a . How Far Are We to GPT-4V? ...

    arXiv 2024

  4. [4]

    Chen, Z.; Wu, J.; Wang, W.; Su, W.; Chen, G.; Xing, S.; Zhong, M.; Zhang, Q.; Zhu, X.; Lu, L.; Li, B.; Luo, P.; Lu, T.; Qiao, Y.; and Dai, J. 2024 b . InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks. In IEEE Conf. Comput. Vis. Pattern Recog., 24185--24198

    work page 2024

  5. [5]

    Cui, Y.; Che, W.; Liu, T.; Qin, B.; and Yang, Z. 2021. Pre-Training With Whole Word Masking for Chinese BERT. IEEE/ACM Trans. Audio, Speech and Lang. Proc., 29: 3504--3514

    work page 2021

  6. [6]

    Dosovitskiy, A.; Beyer, L.; Kolesnikov, A.; Weissenborn, D.; Zhai, X.; Unterthiner, T.; Dehghani, M.; Minderer, M.; Heigold, G.; Gelly, S.; Uszkoreit, J.; and Houlsby, N. 2020. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In Int. Conf. Learn. Represent

    work page 2020

  7. [7]

    Ganin, Y.; Bartunov, S.; Li, Y.; Keller, E.; and Saliceti, S. 2021. Computer-Aided Design as Language. In Adv. Neural Inform. Process. Syst., 5885--5897

    work page 2021

  8. [8]

    Gemini Team . 2024. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context. arXiv:2403.05530

    arXiv 2024

Show all 51 references
  1. [9]

    Gong, J.; Zhang, G.; Zhang, H.; and Sun, J. 2006 a . Reconstruction of 3D curvilinear wire-frame from three orthographic views. Comput. Graph., 30(2): 213--224

    2006

  2. [10]

    Gong, J.; Zhang, H.; Zhang, G.; and Sun, J. 2006 b . Solid reconstruction using recognition of quadric surfaces from orthographic views. Comput. Aided Des., 38(8): 821--835

    2006

  3. [11]

    Gu, K.; Tang, Z.; and Sun, J. 1986. Reconstruction of 3D Objects from Orthographic Projections. Comput. Graph. Forum, 5(4): 317--323

    1986

  4. [12]

    Guo, H.; Liu, S.; Pan, H.; Liu, Y.; Tong, X.; and Guo, B. 2022. ComplexGen : CAD reconstruction by B-rep chain complex generation. ACM Trans. Graph., 41(4): 129:1--129:18

    2022

  5. [13]

    Han, W.; Xiang, S.; Liu, C.; Wang, R.; and Feng, C. 2020. SPARE3D: A Dataset for SPAtial REasoning on Three-View Line Drawings. In IEEE Conf. Comput. Vis. Pattern Recog., 14678--14687

    2020

  6. [14]

    Hu, W.; Zheng, J.; Zhang, Z.; Yuan, X.; Yin, J.; and Zhou, Z. 2023. PlankAssembly: Robust 3D Reconstruction from Three Orthographic Views with Learnt Shape Programs. In IEEE Int. Conf. Comput. Vis., 18495--18505

    2023

  7. [15]

    Idesawa, M. 1973. A System to Generate a Solid Figure from Three View. Bulletin of the JSME , 16(92): 216--225

    1973

  8. [16]

    K.; Lambourne, J

    Jayaraman, P. K.; Lambourne, J. G.; Desai, N.; Willis, K. D. D.; Sanghi, A.; and Morris, N. J. W. 2023. SolidGen : An Autoregressive Model for Direct B-rep Synthesis. Trans. Mach. Learn. Res

    2023

  9. [17]

    K.; Barton, T.; Xu, X.; Wang, K.; Jiang, E.; Guerrero, P.; Mitra, N

    Jones, R. K.; Barton, T.; Xu, X.; Wang, K.; Jiang, E.; Guerrero, P.; Mitra, N. J.; and Ritchie, D. 2020. ShapeAssembly : learning to generate programs for 3D shape structure synthesis. ACM Trans. Graph., 39(6): 234:1--234:20

    2020

  10. [18]

    S.; Dupont, E.; Ali, S

    Khan, M. S.; Dupont, E.; Ali, S. A.; Cherenkova, K.; Kacem, A.; and Aouada, D. 2024. CAD-SIGNet: CAD Language Inference from Point Clouds using Layer-wise Sketch Instance Guided Attention. In IEEE Conf. Comput. Vis. Pattern Recog., 4713--4722

    2024

  11. [19]

    P.; and Ba, J

    Kingma, D. P.; and Ba, J. 2015. Adam: A Method for Stochastic Optimization. In Int. Conf. Learn. Represent

    2015

  12. [20]

    Kuo, M.-H. 1998. Reconstruction of quadric surface solids from three-view engineering drawings. Comput. Aided Des., 30(7): 517--527

    1998

  13. [21]

    Lequette, R. 1988. Automatic construction of curvilinear solids from wireframe views. Comput. Aided Des., 20(4): 171--180

    1988

  14. [22]

    Li, B.; Zhang, K.; Zhang, H.; Guo, D.; Zhang, R.; Li, F.; Zhang, Y.; Liu, Z.; and Li, C. 2024 a . LLaVA-NeXT: Stronger LLMs Supercharge Multimodal Capabilities in the Wild. https://llava-vl.github.io/blog/2024-05-10-llava-next-stronger-llms/

    2024

  15. [23]

    Li, P.; Guo, J.; Li, H.; Benes, B.; and Yan, D.-M. 2024 b . SfmCAD: Unsupervised CAD Reconstruction by Learning Sketch-based Feature Modeling Operations. In IEEE Conf. Comput. Vis. Pattern Recog., 4671--4680

    2024

  16. [24]

    Liu, H.; Li, C.; Li, Y.; and Lee, Y. J. 2024 a . Improved Baselines with Visual Instruction Tuning. In IEEE Conf. Comput. Vis. Pattern Recog., 26296--26306

    2024

  17. [25]

    Liu, H.; Li, C.; Li, Y.; Li, B.; Zhang, Y.; Shen, S.; and Lee, Y. J. 2024 b . LLaVA-NeXT: Improved reasoning, OCR, and world knowledge. https://llava-vl.github.io/blog/2024-01-30-llava-next/

    2024

  18. [26]

    Liu, H.; Li, C.; Wu, Q.; and Lee, Y. J. 2023. Visual Instruction Tuning. In Adv. Neural Inform. Process. Syst., 34892--34916

    2023

  19. [27]

    Liu, S.; Hu, S.; Chen, Y.; and Sun, J. 2001. Reconstruction of curved solids from engineering drawings. Comput. Aided Des., 33(14): 1059--1072

    2001

  20. [28]

    Loshchilov, I.; and Hutter, F. 2017. Decoupled Weight Decay Regularization. In Int. Conf. Learn. Represent

    2017

  21. [29]

    Ma, W.; Chen, S.; Lou, Y.; Li, X.; and Zhou, X. 2024. Draw Step by Step: Reconstructing CAD Construction Sequences from Point Clouds via Multimodal Diffusion. In IEEE Conf. Comput. Vis. Pattern Recog., 27154--27163

    2024

  22. [30]

    Markowsky, G.; and Wesley, M. A. 1980. Fleshing Out Wire Frames. IBM J. Res. Dev. , 24(5): 582--597

    1980

  23. [31]

    Nash, C.; Ganin, Y.; Eslami, S. M. A.; and Battaglia, P. W. 2020. PolyGen : An Autoregressive Generative Model of 3D Meshes. In Int. Conf. Mach. Learn., 7220--7229

    2020

  24. [32]

    OpenAI. 2023. GPT-4V(ision) System Card . https://cdn.openai.com/papers/GPTV_System_Card.pdf

    2023

  25. [33]

    Sakurai, H.; and Gossard, D. C. 1983. Solid Model Input through Orthographic Views. In ACM SIGGRAPH , 243--252

    1983

  26. [34]

    Seff, A.; Zhou, W.; Richardson, N.; and Adams, R. P. 2022. Vitruvion : A Generative Model of Parametric CAD Sketches. In Int. Conf. Learn. Represent

    2022

  27. [35]

    Shin, B.; and Shin, Y. 1998. Fast 3D Solid Model Reconstruction from Orthographic Views. Comput. Aided Des., 30(1): 63--76

    1998

  28. [36]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is All you Need. In Adv. Neural Inform. Process. Syst., 5998--6008

    2017

  29. [37]

    Wang, P.; Bai, S.; Tan, S.; Wang, S.; Fan, Z.; Bai, J.; Chen, K.; Liu, X.; Wang, J.; Ge, W.; Fan, Y.; Dang, K.; Du, M.; Ren, X.; Men, R.; Liu, D.; Zhou, C.; Zhou, J.; and Lin, J. 2024 a . Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution. ar...

    2024 arXiv

  30. [38]

    Wang, W.; and Grinstein, G. G. 1993. A Survey of 3D Solid Reconstruction from 2D Projection Line Drawings. Comput. Graph. Forum, 12(2): 137--158

    1993

  31. [39]

    Wang, W.; Lv, Q.; Yu, W.; Hong, W.; Qi, J.; Wang, Y.; Ji, J.; Yang, Z.; Zhao, L.; Song, X.; Xu, J.; Xu, B.; Li, J.; Dong, Y.; Ding, M.; and Tang, J. 2024 b . CogVLM: Visual Expert for Pretrained Language Models. arXiv:2311.03079

    2024 arXiv

  32. [40]

    A.; and Markowsky, G

    Wesley, M. A.; and Markowsky, G. 1981. Fleshing Out Projections. IBM J. Res. Dev. , 25(6): 934--953

    1981

  33. [41]

    Willis, K. D. D.; Jayaraman, P. K.; Lambourne, J. G.; Chu, H.; and Pu, Y. 2021. Engineering Sketch Generation for Computer-Aided Design. In IEEE Conf. Comput. Vis. Pattern Recog. Worksh., 2105--2114

    2021

  34. [42]

    Wu, K.; Zhang, J.; Peng, H.; Liu, M.; Xiao, B.; Fu, J.; and Yuan, L. 2022. TinyViT: Fast Pretraining Distillation for Small Vision Transformers. In Eur. Conf. Comput. Vis., 68--85

    2022

  35. [43]

    Wu, R.; Xiao, C.; and Zheng, C. 2021. DeepCAD : A Deep Generative Network for Computer-Aided Design Models. IEEE Int. Conf. Comput. Vis., 6772--6782

    2021

  36. [44]

    K.; Lambourne, J

    Xu, X.; Jayaraman, P. K.; Lambourne, J. G.; Willis, K. D. D.; and Furukawa, Y. 2023. Hierarchical Neural Coding for Controllable CAD Model Generation. In Int. Conf. Mach. Learn., 38443--38461

    2023

  37. [45]

    Xu, X.; Willis, K. D. D.; Lambourne, J. G.; Cheng, C.; Jayaraman, P. K.; and Furukawa, Y. 2022. SkexGen: Autoregressive Generation of CAD Construction Sequences with Disentangled Codebooks. In Int. Conf. Mach. Learn., 24698--24724

    2022

  38. [46]

    L.; and Tang, Z

    Yan, Q.; Philip Chen , C. L.; and Tang, Z. 1994. Efficient algorithm for the reconstruction of 3D objects from orthographic projections. Comput. Aided Des., 26(9): 699--717

    1994

  39. [47]

    Yang, A.; Pan, J.; Lin, J.; Men, R.; Zhang, Y.; Zhou, J.; and Zhou, C. 2023. Chinese CLIP: Contrastive Vision-Language Pretraining in Chinese. arXiv:2211.01335

    2023 arXiv

  40. [48]

    You, C.; and Yang, S. 1996. Reconstruction of curvilinear manifold objects from orthographic views. Comput. Graph., 20(2): 275--293

    1996

  41. [49]

    Yu, F.; Chen, Q.; Tanveer, M.; Mahdavi - Amiri, A.; and Zhang, H. 2023. D\( ^ 2 \)CSG: Unsupervised Learning of Compact CSG Trees with Dual Complements and Dropouts. In Adv. Neural Inform. Process. Syst., 22807--22819

    2023

  42. [50]

    Zhao, Y.; Huang, J.; Hu, J.; Zhang, D.; Jiang, Z.; Wu, Z.; Ai, B.; Wang, A.; Zhou, W.; and Chen, Y. 2024. SWIFT: A Scalable lightWeight Infrastructure for Fine-Tuning. arXiv:2408.05517

    2024 arXiv

  43. [51]

    Zhu, D.; Chen, J.; Shen, X.; Li, X.; and Elhoseiny, M. 2024. MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models. In Int. Conf. Learn. Represent

    2024

Pith tools

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