Overview
Creating digital garments from existing two-dimensional patterns remains a time-consuming process. Although patterns are widely available in digital form, their assembly in garment CAD software involves manual definition of sewing operations and arrangement of the pattern panels for drape simulation. This paper presents two learning-based approaches that automate both tasks using the 2D vector pattern as their sole input.
Methodology
The proposed automatic sewing model operates at the segment level, leveraging a Transformer encoder with multiple prediction heads. The novel architecture supports many-to-many, ordered segment correspondences, as well as hems, overlapping seams needed for multilayer garments, and construction properties such as seam allowances and stitch typesβdetails required for the finished virtual garment but neglected by previous methods. A complementary panel-level model predicts body-region assignments, layers, and placement coordinates to initialize the drape simulation.
Evaluation and Results
Both models are evaluated on a curated dataset of 4,035 professionally assembled garment instances. The sewing model exactly recovers all seam and hem topologies of 63.3% of test garments, while the arrangement model meets the specified placement criteria for all panels in 66.2% of test garments. Both methods are also integrated into garment CAD software, demonstrating automatic assembly of successfully predicted garments.
Publication Details
- Authors: Alejandro Rodriguez
- Submitted: 12 October 2026
- arXiv Identifier: arXiv:2610.20651 [cs.CV]
- Subjects: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
- Comments: 11 pages, 8 figures
- DOI: https://doi.org/10.48550/arXiv.2610.20651
Significance
This work addresses a critical bottleneck in digital fashion workflows: the manual effort required to convert 2D patterns into simulation-ready 3D garments. By automating both sewing definition and panel arrangement, the approach reduces the time and expertise needed for digital garment creation. The inclusion of construction properties such as seam allowances and stitch types marks a step beyond prior methods, which typically overlooked these details. As the fashion industry increasingly adopts digital prototyping and virtual try-on technologies, automated assembly pipelines like this are positioned to become essential infrastructure for scalable garment simulation.
via ArXiv CV