초고속영상의 의미론적 분할 및 객체 추적
Semantic Segmentation and Object Tracking of Ultra-high-speed Images
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
초고속영상은 육안으로 판별하기 힘든 현상을 분석할 수 있어 활용도가 높다. 초고속영상에 존재한 객체를 자동인식하기 위하여, 본 논문에서는 객체들을 의미론적 분할하고, 추적하는 방법을 제안한다. 제안하는 방법은 객체분할과 객체추적 2 단계로 구성된다. 객체분할 단계에서는 CNN 기반의 의미론적 분할 모델인 DeepLab v3+을 사용하여 첫 번째 초고속영상에서 있는 관심 객체의 영역을 분할하고, 해당 객체를 추적하기 위하여 경계화소를 추출한다. 객체추적 단계는 두 번째 영상과 그 이후의 영상에 존재한 객체의 영역을 계속적으로 추출하는 단계이다. 제안하는 방법을 성능을 실험 통하여 검증한다. 실험 결과, 객체분할 및 객체추적의 정확도는 각각 약 77%의 mIoU, 약 83%의 mIoU로 평가되었다.
Abstract
Ultra-high speed images are highly utilized as they can analyze the phenomena that are difficult to distinguish with the naked eye. In order to automatically recognize objects existing in the images, this paper proposes a method of semantically segmenting and tracking objects in ultra-high speed images. The proposed method consists of two phases: object segmentation and object tracking. In the object segmentation phase, the method segments an object region in the first ultra-high speed image using DeepLab v3+ which is a kind of CNN-based semantic segmentation models, extract the boundary pixels to track the object. The object tracking phase continuously extracts the object region in the second image and in subsequent images. The performance of the proposed method is verified by experiments. As a result of the experiment, the accuracy of object segmentation and object tracking show about 77% mIoU, about 83% mIoU, respectively.
참고문헌 (13)
- Basilio Pueo, "High speed cameras for motion analy sis in sports science", Journal of Human Sport&Exercise, 11(2), pp.53~73 (2016)
- Jonathan Long, "Fully convolutional networks for semantic segmentation", Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), pp.3431~3440 (2015)
- Seoung Wug Oh, "Video object Segmentation using space-ti me memory networks", ICCV, pp.9226~9235 (2019)
- S. Caelles, "One-shot video objec t segmentation", Proceedings of the IEEE conferenc e on Computer Vision and Pattern Recognition (CV PR), pp.221~230 (2017)
- O. Russakovsky, "ImageNet Large Scale Visual Recognition Challenge", International Journal of Computer Vision, 115(3), pp.211~252 (2015)
- L. C. Chen, "Semantic Image Segmentation wit h Deep Convolutional Nets and Fully Connected CR Fs", pp.1~14 (2015), ICLR
- Liang-Chieh Chen, "Deeplab:Semantic Image Segmentation with Deep Convoluti onal Nets, Atrous Convolution, and Fully Connected CRFs", IEEE Transactions on Pattern Analysis and Machine Intelligence(PAMI), 40(4), pp.834~848 (2018)
- Liang-Chieh Chen, "Rethinking Atrous Convolution for Semantic Image Segmentation", Computer Vision and Pattern Recognition(CVPR), pp.1~14 (2017)
- L. C. Chen, "Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation", pp.1~18 (2018), ECCV
- F. Chollet, "Xception: Deep Learning with Depthwise Separable Convolutions", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.1251~1258 (2017)
- O. Ronneberger, "U-Net : Convolutional Networks for Biomedical Image Seg mentation", International Conference on Medical I mage Computing and Computer-Assisted Intervent ion(MICCAI), pp.234~241 (2015)
- K. He, "Deep Residual Learning for Image Recognition", Proceedings of th e IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp.770~778 (2016)
- J. Bergstra, "Random Search for Hyp er-Parameter Optimization", Journal of Machine Le arning Research, 13, pp.281~305 (2012)