카테터 제조를 위한 토출영역 분할 기반센터링 방향 분류 시스템 = Centering Direction Classification System for Discharge AreaSegmentation based on Catheter Manufacturing

다국어 초록 (Multilingual Abstract)

This paper proposes a vision-based centering direction inference system designed to enhance the accuracy of equipment alignment during the initial setup phase of catheter manufacturing. This system addresses the challenges of centering tasks that are prone to human error and variability, aiming to significantly improve the reliability and efficiency of the production process. Through the construction of a comprehensive catheter centering dataset, this study employs the advanced capabilities of the SAM2 (Segment Anything Model 2) to extract precise mask data from catheter material extrusion sequences. These masks provide detailed insights into the positional changes from the initial to the final frame, which are essential for inferring the accurate centering direction. The algorithm developed categorizes the centering direction into three distinct classes: ‘Normal’, ‘Right’, and ‘Left’. It demonstrates robust performance with an accuracy of approximately 88.8% showcasing the system’s effectiveness across different material types. This level of accuracy is crucial for ensuring the quality and consistency of catheter products. Moreover, the paper discusses the potential for future research to expand the application of the SAM2 algorithm to enhance centering precision for a broader range of materials and catheter shapes. The ongoing development of this technology is expected to further automate and refine manufacturing processes, pushing the boundaries of what is currently achievable in high-precision manufacturing environments. The implementation of such advanced manufacturing technologies not only streamlines production but also facilitates a shift towards smarter manufacturing practices.

참고문헌 (Reference)

1 H. Zhang, "Self-attention Generative Adversarial Networks" PMLR 7354-7363, 2019

2 A. Kirillov, "Segment Anything" 4015-4026, 2023

3 N. Ravi, "Sam 2: Segment Anything in Images and Videos"

4 S. H. Lee, "Predicting the Heading Angle of Resin During Extrusion Using Semantic Segmentation Based on Edge-region Focal Loss" 2024

5 H. Li, "PRISM: A Promptable and Robust Interactive Segmentation Model with Visual Prompts" Springer Nature Switzerland 389-399, 2024

6 R. Bommasani, "On the Opportunities and Risks of Foundation Models"

7 K. He, "Masked Autoencoders are Scalable Vision Learners" 16000-16009, 2022

8 W. A. Hyman, "Manufacturing Defects" 49 (49): 81-83, 2019

9 O. O. Vergara-Villegas, "Lean Manufacturing in the Developing World: Methodology, Case Studies and Trends from Latin America" 135-156, 2014

10 F. A. Lievano-Martínez, "Intelligent Process Automation : An Application in Manufacturing Industry" 14 (14): 8804-, 2022

11 M. Bramson, "Instability of FIFO Queueing Networks" 4 (4): 414-431, 1994

12 M. Bożek, "Improvement of Catheter Quality Inspection Process" 121 (121): 2017

13 C. Ryali, "Hiera: A Hierarchical Vision Transformer Without the Bells-and-whistles" PMLR 29441-29454, 2023

14 A. Fitzgibbon, "Direct Least Square Fitting of Ellipses" 21 (21): 476-480, 1999

15 S. H. Jeong, "Development of Microcatheter Tube Extrusion Angle Estimation System Using Convolutional Neural Network Segmentation" 13 (13): 18468-, 2023

16 M. Jolaei, "Design, Development and Force Control of a Tendon-driven Steerable Catheter with a Learning-based Approach" Concordia University 2020

17 Z. Huang, "Ccnet : Criss-cross Attention for Semantic Segmentation" 603-612, 2019

18 R. Cioffi, "Artificial Intelligence and Machine Learning Applications in Smart Production : Progress, Trends, and Directions" 12 (12): 492-, 2020

19 J. Xu, "A Review on AI for Smart Manufacturing : Deep Learning Challenges and Solutions" 12 (12): 8239-, 2022

20 Y. Xie, "A New Efficient Ellipse Detection Method" IEEE 2 : 957-960, 2002