This study proposes an intelligent system for automatically generating men’s suit patterns, aiming to streamline the pattern-making process with improved automation and efficiency. By employing an image recognition model integrated with computer-aided design (CAD) and parameterized design principles, the study addresses the challenges of traditional manual methods that are time-consuming and highly dependent on designers’ expertise.
An enhanced YOLOv8-pose model is employed for keypoint detection, enabling the identification of critical suit features from images. The extracted keypoints are then mapped to actual suit dimensions using linear regression models, establishing a consistent proportionality framework for accurate pattern scaling. Custom pattern generation rules were developed and implemented in AutoCAD via Python, facilitating automatic drafting and dynamic adjustments. A comparative experiment was conducted, comparing automated patterns with those created by traditional methods to validate accuracy.
The system demonstrates significant improvements in pattern-making efficiency and accuracy, with keypoint detection and automated drafting consistently aligning with traditional hand-crafted patterns. The automated system proved to be capable of generating patterns with less than 0.5 cm deviation in key dimensions, verifying the system’s reliability for accurate, scalable and customizable suit pattern generation.
This research integrates state-of-the-art image recognition with CAD systems in men’s suit design, advancing traditional pattern-making by introducing an automated, adaptive approach that can meet the demands of personalized fashion.
