반도체디스플레이기술학회지 (Journal of the Semiconductor & Display Technology)

한국반도체디스플레이기술학회 (The Korean Society Of Semiconductor & Display Technology)
권호 표지

인공지지체 불량 검출을 위한 딥러닝 모델 손실 함수의 성능 비교

Performance Comparison of Deep Learning Model Loss Function for Scaffold Defect Detection

  • 이송연 (한국기술교육대학교 인공지능연구실) ;
  • 허용정 (한국기술교육대학교 메카트로닉스공학부)
  • Song Yeon Lee (Artificial Intelligence Laboratory, Korea University of Technology and Education) ;
  • Yong Jeong Huh (School of Mechatronics Engineering, Korea University of Technology and Education)
  • 투고 : 2023.05.30
  • 심사 : 2023.06.21
  • 발행 : 2023.06.30
PDF 다운로드
⟨ 이전 논문 다음 논문 ⟩

초록

The defect detection based on deep learning requires minimal loss and high accuracy to pinpoint product defects. In this paper, we confirm the loss rate of deep learning training based on disc-shaped artificial scaffold images. It is intended to compare the performance of Cross-Entropy functions used in object detection algorithms. The model was constructed using normal, defective artificial scaffold images and category cross entropy and sparse category cross entropy. The data was repeatedly learned five times using each loss function. The average loss rate, average accuracy, final loss rate, and final accuracy according to the loss function were confirmed.

키워드

과제정보

이 논문은 2023년도 한국기술교육대학교 교수 교육연구진흥과제 지원에 의하여 연구되었음.

참고문헌

  1. Song-Yeon Lee and Yong-Jeong Huh, "A Study on Shape Warpage Defect Detection Medel of Scaffold Using Deep Learning Base CNN", J. of The Korean Society of Semiconductor & Display Technology, Vol.20, pp. 99-103, 2021
  2. Song-Yeon Lee and Yong-Jeong Huh, "A Study on Surface Defect Detection Model of 3D Printing Bone Plate Using Deep Learning Algorithm", J. of The Korean Society of Semiconductor & Display Technology, Vol.20, pp. 68-73, 2021
  3. Jun-Hee Jung and Joong-Hwee Cho, "A Through-focus Scanning Optical Microscopy Dimensional Measurement Method based on A Deep-learning Regression Model", J. of The Korean Society of Semiconductor & Display Technology, Vol.21, pp. 108-113, 2022
  4. Gee-Yeun Kim and Hyoung-Gook Kim, "Performanc Comparison of Lung Sound Classification Using Various Convolutional Neural Networks," J. of The Acoustical Society of Korea, Vol. 38, pp. 568-573, 2019
  5. Chang-Hee Yang, Kyu-Sub Park, Young-Seop Kim and Yong-Hwan Lee, "Comparative Analysis for Emotion Expression Using Three Methods Based by CNN", J. of The Korean Society of Semiconductor & Display Technology, Vol.19, pp. 65-70, 2020
  6. Ji-Soo Kang, Se-Eun Shim, Sun-Moon Jo and Kyung- Yong Chung, "YOLO based Light Source Object Detection for Traffic Image Big Data Processing", J. of The Korean Convergence for Information Technology, Vol.10, pp. 40-46, 2020
  7. Se-Rang Oh and Young-Chul Bae, "Braille Block Recognition Algorithm for the Visually Impaired Based on YOLO V3", J. of The Korean Institute of Intelligent Systems, Vol.31, pp. 60-67, 2021 https://doi.org/10.5391/JKIIS.2021.31.1.060
  8. Hyun-Jun Kim and Dong-Wan Choi, "A Perimeter-Based IoU Loss for Efficient Bounding Box Regression in Object Detection", J of The Korean Institute of Information Scientists and Engineers, Vol.48, pp.913-919, 2021. https://doi.org/10.5626/JOK.2021.48.8.913