An Armature Defect Self-Adaptation Quantitative Assessment System Based on Improved YOLO11 and the Segment Anything Model

Abstract

There is a need to address challenges faced in detecting and segmenting defects in micro-vibration motor armatures, which are crucial components used in digital devices. Due to their complex structure and tiny size, quality control during assembly is difficult. In this paper, an adaptive segmentation quantization (ASQ) system based on YOLO 11 and SAM is proposed to address the issue above. The system consists of a target detection (TD) unit, shape segmentation (SS) unit, and quantitative assessment (AS) unit, and introduces a practical combination of YOLO11 for defect detection and SAM for segmentation, integrating this with a novel quantitative assessment framework to measure defect severity and occurrence. This approach is efficient and cost-effective, supporting real-time industrial applications by allowing for automated, rapid analysis and improvement identification. Finally, a quantitative evaluation standard with more than 90% accuracy was achieved. Additionally, a hardware system was developed to implement this framework in industrial settings. The proposed framework adopts a strategy of intelligent morphological feature extraction and computation, focusing on pixel-level segmentation and quantitative assessment. This research makes a significant step forward in automating quality control processes for micro-scale components, providing a robust and adaptive solution for the enhancement of manufacturing efficiency and product quality.