Buckets:
| license: apache-2.0 | |
| pretty_name: Cadsy CAD Technical Drawings | |
| tags: | |
| - cad | |
| - mechanical-cad | |
| - step | |
| - technical-drawing | |
| - engineering-drawing | |
| - iso | |
| - asme | |
| - multimodal | |
| - synthetic-data | |
| configs: | |
| - config_name: default | |
| data_files: | |
| - split: train | |
| path: metadata.jsonl | |
| # CAD Technical Drawings, Generated by Cadsy | |
| Turn a STEP model into a labeled technical drawing automatically. | |
| This sample was created with **Cadsy** from 3D models in the | |
| [Zero-to-CAD-100k dataset](https://huggingface.co/datasets/ADSKAILab/Zero-To-CAD-100k). | |
| For every STEP model, Cadsy generated: | |
| - one drawing using an ASME-style profile; | |
| - one drawing using an ISO-style profile; and | |
| - structured bounding-box labels for every retained annotation. | |
| That is **65 CAD models, 130 technical drawings and their labels**, produced | |
| through one repeatable generation pipeline. | |
| ## Example | |
|  | |
| An ISO-style drawing generated from a STEP model, with the annotation | |
| boxes from its labels file drawn on top of the image. | |
| ## Build CAD drawing datasets without drawing every sheet by hand | |
| Cadsy turns existing CAD geometry into consistent, machine-readable drawing | |
| data. It can help teams create training sets, evaluation sets and | |
| customer-specific datasets for systems that need to understand mechanical CAD | |
| and technical drawings. | |
| Use Cadsy to generate: | |
| - paired STEP models and technical drawings; | |
| - controlled ISO-style and ASME-style variants; | |
| - annotation locations and text labels; | |
| - repeatable datasets at a larger scale; and | |
| - data tailored to particular part families or research tasks. | |
| This repository is a public sample. To generate your own dataset with an API, visit **[cadsy.se](https://cadsy.se)**. | |
| To provide any feedback please wirte to **hello@cadsy.se**. | |
| ## Dataset contents | |
| ```text | |
| metadata.jsonl | |
| samples/ | |
| part_0001/ | |
| model.step # source CAD geometry | |
| asme.png # drawing generated with the ASME-style profile | |
| asme_labels.json # annotations for asme.png | |
| iso.png # drawing generated with the ISO-style profile | |
| iso_labels.json # annotations for iso.png | |
| part_0002/ | |
| ... | |
| ``` | |
| `metadata.jsonl` contains one line per part: | |
| | Field | Type | Description | | |
| |---|---|---| | |
| | `part_id` | string | Part identifier, such as `"part_0001"` | | |
| | `source_row` | int | Row index in the upstream Zero-to-CAD-100k dataset | | |
| | `uuid` | string | UUID from the upstream row | | |
| | `model_step` | string | Relative path to the STEP file | | |
| | `asme_image` | string | Relative path to the ASME-style drawing | | |
| | `asme_labels` | string | Relative path to the ASME-style drawing labels | | |
| | `iso_image` | string | Relative path to the ISO-style drawing | | |
| | `iso_labels` | string | Relative path to the ISO-style drawing labels | | |
| Each `*_labels.json` file contains `{"annotations": [...]}`. Every annotation | |
| has: | |
| - `type` — `label`, `dimension`, `callout`, `feature_control_frame`, | |
| `datum_definition`, `title_block_value`, `title_block_label`, or | |
| `projection_third_angle`; | |
| - `view` — `sheet`, `front`, `side`, or `plan`; | |
| - `text` — the annotation text; and | |
| - `box` — `[x0, y0, x1, y1]`, normalized to the 2480 × 1754 image. | |
| ## Important note | |
| The ISO-style and ASME-style outputs are controlled synthetic generation | |
| profiles. They are intended for dataset development and evaluation; they are | |
| not certified manufacturing drawings or a substitute for engineering review. | |
| ## License | |
| This sample dataset is released under Apache License 2.0. See `LICENSE`. See [Zero-to-CAD-100k dataset](https://huggingface.co/datasets/ADSKAILab/Zero-To-CAD-100k) | |
| for source license. | |
Xet Storage Details
- Size:
- 3.71 kB
- Xet hash:
- 9863cf06b1a5fd18fe6853eec843d1ad2a7dff3d22c5e8ac82290cff23d86941
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.