로봇학회논문지 제14권 제1호 (통권 제51호) (p.14-21)

수중 소나 영상 학습 데이터의 왜곡 및 회전 Augmentation을 통한 딥러닝 기반의 마커 검출 성능에 관한 연구

Study of Marker Detection Performance on Deep Learning via Distortion and Rotation Augmentation of Training Data on Underwater Sonar Image
키워드 :
Deep Learning,Data Augmentation,Object Detection,Underwater Sonar Image

목차

Abstract
1. 서 론
2. 관련 연구
  2.1 수중 초음파센서
  2.2 물체 감지
3. 데이터 Augmentation
  3.1 학습 데이터
  3.2 테스트 데이터
4. 실험 결과
  4.1 Rotation Augmentation
  4.2 테스트 데이터의 세트화
5. 결 론
References

초록

In the ground environment, mobile robot research uses sensors such as GPS and optical cameras to localize surrounding landmarks and to estimate the position of the robot. However, an underwater environment restricts the use of sensors such as optical cameras and GPS. Also, unlike the ground environment, it is difficult to make a continuous observation of landmarks for location estimation. So, in underwater research, artificial markers are installed to generate a strong and lasting landmark. When artificial markers are acquired with an underwater sonar sensor, different types of noise are caused in the underwater sonar image. This noise is one of the factors that reduces object detection performance. This paper aims to improve object detection performance through distortion and rotation augmentation of training data. Object detection is detected using a Faster R-CNN.