Abstract:We introduce CADSpotting, an efficient method for panoptic symbol spotting in large-scale architectural CAD drawings. Existing approaches struggle with the diversity of symbols, scale variations, and overlapping elements in CAD designs. CADSpotting overcomes these challenges by representing each primitive with dense points instead of a single primitive point, described by essential attributes like coordinates and color. Building upon a unified 3D point cloud model for joint semantic, instance, and panoptic segmentation, CADSpotting learns robust feature representations. To enable accurate segmentation in large, complex drawings, we further propose a novel Sliding Window Aggregation (SWA) technique, combining weighted voting and Non-Maximum Suppression (NMS). Moreover, we introduce a large-scale CAD dataset named LS-CAD to support our experiments. Each floorplan in LS-CAD has an average coverage of 1,000 square meter(versus 100 square meter in the existing dataset), providing a valuable benchmark for symbol spotting research. Experimental results on FloorPlanCAD and LS-CAD datasets demonstrate that CADSpotting outperforms existing methods, showcasing its robustness and scalability for real-world CAD applications.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve the problem of panoptic symbol spotting in large - scale architectural CAD drawings. Specifically, existing methods face the following challenges when dealing with symbol recognition in CAD drawings: 1. **Symbol diversity**: There are a large variety of symbols in CAD drawings, and there may be visual similarities between different symbols. 2. **Scale variation**: The size and proportion of symbols vary greatly in different drawings. 3. **Element overlap**: Symbols and other graphic elements in CAD drawings may overlap each other, increasing the difficulty of recognition. To solve these problems, the author proposes a new method named **CADSpotting**. This method improves the effect of symbol recognition in the following aspects: - **Dense point sampling**: Represent CAD graphic primitives as densely sampled point clouds, where each point contains coordinate and color information, thus constructing a richer feature representation. - **Unified 3D point - cloud model**: Use a unified 3D point - cloud processing model to learn robust feature representations and support semantic, instance, and panoptic segmentation tasks. - **Sliding Window Aggregation (SWA) technique**: Introduce the SWA technique during the inference process, combined with weighted voting and non - maximum suppression (NMS), to improve the accuracy and efficiency of symbol recognition in large - scale CAD drawings. In addition, the author also introduces a new large - scale CAD dataset **LS - CAD**, which contains 50 annotated floor plans with an average coverage area of more than 1,000 square meters, providing a valuable benchmark for symbol recognition research. ### Formula presentation Some of the formulas involved in the paper are as follows: - Six - dimensional vector representation of each point: \[ f=(x, y, z, r, g, b)\in\mathbb{R}^6 \] where \(x, y\) represent the two - dimensional coordinates of the point, \(z\) is set to 0, and \(r, g, b\) represent the color information of the point (red, green, blue channels) respectively. - Conversion from original point features to primitive - level features: \[ f\in\mathbb{R}^{M\times C}\to g\in\mathbb{R}^{N\times C} \] where \(M\) represents the total number of 2D sampling points and \(N\) represents the number of primitives. - Calculation of primitive - level features: \[ g_i(c)=\max_{p\in P_i}f_p(c)+\frac{1}{|P_i|}\sum_{p\in P_i}f_p(c) \] - Loss function: \[ L = \lambda_{\text{cls}}L_{\text{cls}}+\lambda_{\text{bce}}L_{\text{bce}}+\lambda_{\text{dice}}L_{\text{dice}} \] Through these methods and techniques, CADSpotting can more efficiently handle the symbol recognition problem in large - scale CAD drawings and has achieved better results than existing methods on multiple datasets.

CADSpotting: Robust Panoptic Symbol Spotting on Large-Scale CAD Drawings

FloorPlanCAD: A Large-Scale CAD Drawing Dataset for Panoptic Symbol Spotting

Symbol as Points: Panoptic Symbol Spotting via Point-based Representation

Pixel-Wise Symbol Spotting via Progressive Points Location for Parsing CAD Images

Symbol Spotting on Digital Architectural Floor Plans Using a Deep Learning-based Framework

Panoptic Segmentation for Seasonal River Scene Based on Spatial Context Prior and DenseASPP

SymPoint Revolutionized: Boosting Panoptic Symbol Spotting with Layer Feature Enhancement

Symbol spotting for architectural drawings: state-of-the-art and new industry-driven developments

An Example-Driven Symbol Recognition Approach Based on Key Features in Engineering Drawings

Point2CAD: Reverse Engineering CAD Models from 3D Point Clouds

A Structural Method for Online Sketched Symbol Recognition.

Research on automated recognizing and interpreting architectural drawings

An Improved Example-Driven Symbol Recognition Approach in Engineering Drawings.

Complex CAD Surface Shape Design Using Semantic Features

The Computer Vision-based Tolerancing Callout Detection Model

Few-Shot Symbol Detection in Engineering Drawings

CAD-SIGNet: CAD Language Inference from Point Clouds using Layer-wise Sketch Instance Guided Attention

Automatic analysis and integration of architectural drawings

CAD-Estate: Large-scale CAD Model Annotation in RGB Videos

Mask2CAD: 3D Shape Prediction by Learning to Segment and Retrieve

Leveraging Deep Convolutional Neural Network for Point Symbol Recognition in Scanned Topographic Maps