ArchCAD-400K: A Large-Scale CAD drawings Dataset and New Baseline for Panoptic Symbol Spotting

TL;DR

This work tackles panoptic symbol spotting in architectural CAD drawings by introducing ArchCAD-400k, a large-scale, richly annotated vector dataset created via an efficient layer-block–driven annotation pipeline, and a strong baseline model, DPSS, that fuses image and geometric primitive features with an adaptive fusion module. DPSS demonstrates state-of-the-art performance on FloorPlanCAD and ArchCAD-400k, highlighting improved robustness and scalability on diverse, large-scale CAD drawings. The dataset expands diversity across building types, spatial scales, and 27 semantic categories, enabling broader AI-enabled architectural design and construction applications. Together, ArchCAD-400k and DPSS advance CAD symbol understanding by delivering scalable annotation, richer semantics, and robust panoptic segmentation without heavy reliance on priors.

Abstract

Recognizing symbols in architectural CAD drawings is critical for various advanced engineering applications. In this paper, we propose a novel CAD data annotation engine that leverages intrinsic attributes from systematically archived CAD drawings to automatically generate high-quality annotations, thus significantly reducing manual labeling efforts. Utilizing this engine, we construct ArchCAD-400K, a large-scale CAD dataset consisting of 413,062 chunks from 5538 highly standardized drawings, making it over 26 times larger than the largest existing CAD dataset. ArchCAD-400K boasts an extended drawing diversity and broader categories, offering line-grained annotations. Furthermore, we present a new baseline model for panoptic symbol spotting, termed Dual-Pathway Symbol Spotter (DPSS). It incorporates an adaptive fusion module to enhance primitive features with complementary image features, achieving state-of-the-art performance and enhanced robustness. Extensive experiments validate the effectiveness of DPSS, demonstrating the value of ArchCAD-400K and its potential to drive innovation in architectural design and construction.

Figures (17)

• Figure 1: Raw CAD Drawings
• Figure 2: Panoptic labeled Drawings
• Figure 4:
• Figure 5:
• Figure 7: Overall pipeline of the annotation process. Primitives are labeled by automated methods followed by manual modification.