RealCAD: Towards Real-World Image-to-CAD Reconstruction under Domain Shift and Parameter Bias
2026-08-31 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors focus on improving how 3D design models (CAD) are made from images, especially real photos instead of just computer-generated ones. They found that previous methods relied too much on common number patterns rather than truly understanding the shapes from images. To fix this, they changed how design details are represented, made synthetic images look more like real ones, and improved how features from different views are compared. They also created a new dataset of real photos with exact design steps. Their approach helps better predict and edit CAD models from real pictures.
Computer-Aided Design (CAD)Image-to-CAD reconstructionDomain gapParameter biasRepresentation learningContrastive learningSynthetic renderingGeometry-constrained translation3D-printed objects datasetMulti-view imaging
Authors
Yihe Sun, Ziyu Lu, Kaihua Tang, Xian-Sheng Hua
Abstract
Reconstructing editable Computer-Aided Design (CAD) models from images is essential for downstream modification, manufacturing, and design reuse. However, existing image-to-CAD methods are developed predominantly on synthetic renderings and face two coupled obstacles: a substantial appearance domain gap between synthetic and real images, and a previously overlooked parameter bias in widely used CAD data. We show that the local normalization adopted by DeepCAD concentrates several geometric parameters around a few discrete values while encoding substantial information in a single scale factor. Consequently, a model can achieve deceptively high parameter accuracy by exploiting these frequent values rather than inferring geometry from the input image. In this paper, we propose RealCAD, a unified framework that addresses these limitations at the representation, image, and feature levels. At the representation level, we redistribute scale information to the corresponding geometric parameters, producing less concentrated parameter distributions in a shared scale space. At the image level, geometry-constrained translation converts synthetic renderings toward the real-image domain while conditioning on object contours. At the feature level, a multi-positive contrastive objective aligns representations of the same CAD model across viewpoints and image domains, enabling CAD sequence prediction from each individual view. We further introduce OpenRealCAD, comprising four-view photographs of 392 3D-printed objects paired with ground-truth command sequences. Experiments show that the revised representation substantially reduces the accuracy attainable from parameter-frequency priors, making parameter accuracy a more reliable measure of image-conditioned geometric inference. RealCAD further improves real-domain command and parameter accuracy, while retaining competitive synthetic-domain performance.