--- license: cc-by-nc-4.0 task_categories: - visual-question-answering - image-to-text viewer: false # Gated form configuration extra_gated_fields: full_name: type: "text" label: "Full Name" affiliation: type: "text" label: "Affiliation / Organization" email: type: "text" label: "Institutional Email (preferred)" country: type: "country" label: "Country / Region" intended_use: type: "select" label: "Intended Use Case" options: ["Academic research", "Educational purpose", "Internal enterprise testing (non-commercial)", "Other"] non_commercial_use: type: "checkbox" label: "I confirm that this dataset will be used for non-commercial purposes only" data_protection_ack: type: "checkbox" label: "I confirm that I will not attempt to extract personal, sensitive, or confidential information from this dataset" extra_gated_prompt: | To request access to the **ArchCAD** dataset, please complete the following form. Access is restricted to non-commercial use. All requests will be reviewed, and approval may take up to 3 business days. --- # πŸ—οΈ ArchCAD [πŸ‡ΊπŸ‡Έ English](README.md) | [πŸ‡¨πŸ‡³ δΈ­ζ–‡θ―΄ζ˜Ž](README_zh-CN.md)
[![arXiv](https://img.shields.io/badge/arXiv-2503.22346-b31b1b.svg)](https://arxiv.org/abs/2503.22346) [![Code](https://img.shields.io/badge/Code-Github-blue)](https://github.com/ArchiAI-LAB/ArchCAD) [![Dataset](https://img.shields.io/badge/Dataset-HuggingFace-yellow)](https://huggingface.co/datasets/jackluoluo/ArchCAD)
*A Multimodal CAD Dataset for Vectorized Drawing Understanding* > **40k Samples Β· 5 Strictly Aligned Modalities Β· Foundational Data for AI Understanding of Engineering Drawings** --- ## πŸ“‘ Table of Contents - [What is ArchCAD?](#πŸ“˜-what-is-archcad) - [Key Features](#✨-key-features) - [Dataset Structure](#🧩-dataset-structure) - [Data Modalities](#data-modalities) - [Annotations](#annotations) - [Baseline Model: DPSS](#πŸ†-baseline-model-dpss) - [Potential Applications](#πŸš€-potential-applications) - [Citation](#πŸ“š-citation) --- ## πŸ“˜ What is ArchCAD? AI systems have long struggled to interpret and utilize CAD drawings directly β€” a major bottleneck in automation for the construction industry. **ArchCAD** aims to address this challenge by providing a large-scale, high-quality CAD dataset with fine-grained, primitive-level annotations. It serves as a foundation for training AI models with deep multimodal understanding of architectural and engineering drawings. Each sample is generated through a **β€œdrawing slicing”** process, where each slice corresponds to a 14m Γ— 14m area of a floor plan, aligned across multiple data modalities. --- ## ✨ Key Features - **🧠 Rich Multimodal Data**: Each sample includes five synchronized data types β€” raster image, SVG, JSON, Q&A, and point cloud. - **πŸ›οΈ Primitive-Level Annotation**: Every geometric entity (e.g., line, arc, circle) within a slice carries both semantic and instance-level labels. - **🧩 Strict Data Alignment**: All five modalities correspond exactly by filename. --- ## 🧩 Dataset Structure ### Data Modalities
Each ArchCAD sample provides five data modalities: 1. **Rasterized Images**: Standard pixel representation of CAD drawings. 2. **SVG Data (Vector Graphics)**: Preserves full vector information. ```xml ... ``` 3. **JSON Data**: Explicit representation of every geometric element as an independent object with all properties stored as key-value pairs. ```json [ { "type": "LINE", "linetype": "Continuous", "start": [x1, y1], "end": [x2, y2], "rgb": [0, 0, 0], "semantic": "single_door", "instance": "single_door_23" }, { "type": "CIRCLE", "center": [x1, y1], "radius": r1, "semantic": "single_door", "instance": "single_door_23" } ] ``` 4. **Q&A Pairs**: Automatically generated structured question-answer pairs for each drawing slice. - **question:** How many doors are in the drawing? **answer:** There are 11 doors in total, including 7 single doors, 1 double door, 1 parent-child door, and 2 other types. - **question:** What are the coordinates of the staircase? **answer:** The bounding box of the staircase has vertices at (0, 624), (104, 624), (104, 827), (0, 827). 5. **Point Cloud Data**: Key control points from geometric primitives (e.g., lines, arcs, circles) represented as 2D point clouds with class labels. ### Annotations Each geometric primitive in ArchCAD carries semantic information, and countable entities additionally have instance-level labels: - `semantic`: the semantic class of the primitive (mapping provided below). - `instance`: a unique identifier shared by all primitives belonging to the same object instance. | ID | Name | Countable | |-----|--------------------|------------| | 0 | Axis & Grid | No | | 1 | Single Door | Yes | | 2 | Double Door | Yes | | 3 | Parent-Child Door | Yes | | 4 | Other Door | Yes | | 5 | Elevator | Yes | | 6 | Staircase | Yes | | 7 | Sink | Yes | | 8 | Urinal | Yes | | 9 | Toilet | Yes | | 10 | Bathtub | Yes | | 11 | Squat Toilet | Yes | | 12 | Other Fixtures | No | | 13 | Drain | No | | 14 | Table | Yes | | 15 | Chair | Yes | | 16 | Bed | Yes | | 17 | Sofa | Yes | | 18 | Hole | Yes | | 19 | Glass | No | | 20 | Wall | No | | 21 | Concrete Column | Yes | | 22 | Steel Column | Yes | | 23 | Concrete Beam | No | | 24 | Steel Beam | No | | 25 | Parking Space | Yes | | 26 | Foundation | No | | 27 | Pile | Yes | | 28 | Rebar | No | | 29 | Fire Hydrant | Yes | | 100 | Others | No | --- ## πŸ“š Citation If you use ArchCAD in your research, please cite the following paper: ```bibtex @article{Luo2025ArchCAD, title={ArchCAD-400K: An Open Large-Scale Architectural CAD Dataset and New Baseline for Panoptic Symbol Spotting}, author={Luo R, Liu Z, Cheng T, et al.}, journal={arXiv preprint arXiv:2503.22346}, year={2025} } ```