딥러닝 기반 연기 합성 기법을 최적화하기 위한 다중 이진화 쿼드트리 프레임워크
Multiple Binarization Quadtree Framework for Optimizing Deep Learning-Based Smoke Synthesis Method
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 논문에서는 초해상도(Super-Resolution, SR)을 계산하는데 필요한 물리 기반 시뮬레이션 데이터를 효율적으로 분류하고 분할하여 빠르게 SR연산을 가능하게 하는 쿼드트리 기반 최적화 기법을 제안한다. 제안하는 방법은 입력 데이터로 사용하는 연기 시뮬레이션 데이터를 다운스케일링(Downscaling)하여 쿼드트리 연산 소요 시간을 대폭 감소시킨다. 이 과정에서 연기의 밀도를 이진화함으로써, 다운스케일링 과정에서 밀도가 수치 손실되는 문제를 완화하며 쿼드트리를 구축한다. 학습에 사용된 데이터는 COCO 2017 데이터 셋이며, 인공신경망은 VGG19 기반 네트워크를 사용한다. 컨볼루션 계층을 거칠 때 데이터의 손실을 막기 위해 잔차(Residual) 보완 방식과 유사하게이전 계층의 출력 값을 더해주며 학습을 진행한다. 실험결과가 연기의 경우 제안된 방법은 이전접근법에 비해 약 15~18배 정도의 속도향상을 얻었다.
Abstract
In this paper, we propose a quadtree-based optimization technique that enables fast Super-resolution(SR) computation by efficiently classifying and dividing physics-based simulation data required to calculate SR. The proposed method reduces the time required for quadtree computation by downscaling the smoke simulation data used as input data. By binarizing the density of the smoke in this process, a quadtree is constructed while mitigating the problem of numerical loss of density in the downscaling process. The data used for training is the COCO 2017 Dataset, and the artificial neural network uses a VGG19-based network. In order to prevent data loss when passing through the convolutional layer, similar to the residual method, the output value of the previous layer is added and learned. In the case of smoke, the proposed method achieved a speed improvement of about 15 to 18 times compared to the previous approach.
참고문헌 (28)
- Park H, "Understanding the Stability of Deep Control Policies for Biped Locomotion", arXiv preprint arXiv:2007.15242, ~ (2020)
- Jungdam Won, "Aerobatics control of flying creatures via self-regulated learning", ACM Transactions on Graphics, 37(6), pp.1~10 (2018), Association for Computing Machinery (ACM)[DOI]
- Lee, Jaedong, "Crowd simulation by deep reinforcement learning", Proceedings of the 11th Annual International Conference on Motion, Interaction, and Games, pp.1~7 (2018)
- Tompson, Jonathan, "Accelerating eulerian fluid simulation with convolutional networks", International Conference on Machine Learning, pp.3424~3433 (2017), PMLR
- Cheng Yang, "Data-driven projection method in fluid simulation", Computer Animation and Virtual Worlds, 27(3-4), pp.415~424 (2016), Wiley[DOI]
- Byungsoo Kim, "Deep Fluids: A Generative Network for Parameterized Fluid Simulations", Computer Graphics Forum, 38(2), pp.59~70 (2019), Wiley[DOI]
- Yongcheng Jing, "Neural Style Transfer: A Review", IEEE Transactions on Visualization and Computer Graphics, 26(11), pp.3365~3385 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Azadi, Samaneh, "Multi-content gan for few-shot font style transfer", Proceedings of the IEEE conference on computer vision and pattern recognition, pp.7564~7573 (2018)
- Yu-shan Sun, "Design of motion control of dam safety inspection underwater vehicle", Journal of Central South University, 19(6), pp.1522~1529 (2012), Springer Science and Business Media LLC[DOI]
- Masaki Oshita, "A Dynamic Motion Control Technique for Human-like Articulated Figures", Computer Graphics Forum, 20(3), pp.192~203 (2001), Wiley[DOI]
- Xingtong Liu, "Numerical simulation and experimental investigation of GaN-based flip-chip LEDs and top-emitting LEDs", Applied Optics, 56(34), pp.9502~ (2017), The Optical Society[DOI]
- 홍병선, "CNN을 이용한 Quad Tree 기반 2D Smoke Super-resolution", 한국컴퓨터그래픽스학회논문지, 25(3), pp.105~113 (2019), (사)한국컴퓨터그래픽스학회[DOI]
- Theodore Kim, "Wavelet turbulence for fluid simulation", ACM Transactions on Graphics, 27(3), pp.1~6 (2008), Association for Computing Machinery (ACM)[DOI]
- Adam W. Bargteil, "A semi-Lagrangian contouring method for fluid simulation", ACM Transactions on Graphics, 25(1), pp.19~38 (2006), Association for Computing Machinery (ACM)[DOI]
- Shah, Maurya, "Extended galilean invariance for adaptive fluid simulation", Proceedings of the 2004 ACM SIGGRAPH, 2004, 213~221
- Rajsekhar Setaluri, "SPGrid", ACM Transactions on Graphics, 33(6), pp.1~12 (2014), Association for Computing Machinery (ACM)[DOI]
- Frank Losasso, "Simulating water and smoke with an octree data structure", ACM Transactions on Graphics, 23(3), pp.457~462 (2004), Association for Computing Machinery (ACM)[DOI]
- McAdams, Aleka, "A Parallel Multigrid Poisson Solver for Fluids Simulation on Large Grids", Symposium on Computer Animation, pp.65~73 (2010)
- Xiangyun Xiao, "A Novel CNN-Based Poisson Solver for Fluid Simulation", IEEE Transactions on Visualization and Computer Graphics, 26(3), pp.1454~1465 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Mengyu Chu, "Data-driven synthesis of smoke flows with CNN-based feature descriptors", ACM Transactions on Graphics, 36(4), pp.1~14 (2017), Association for Computing Machinery (ACM)[DOI]
- Li, Jianan, "Perceptual generative adversarial networks for small object detection", Proceedings of the IEEE conference on computer vision and pattern recognition, pp.1222~1230 (2017)
- Huang, Wenqing, "Detection of traffic signs based on combination of GAN and faster-RCNN", Journal of Physics: Conference Series, 1069(1), pp.012159~ (2018)[DOI]
- You Xie, "tempoGAN", ACM Transactions on Graphics, 37(4), pp.1~15 (2018), Association for Computing Machinery (ACM)[DOI]
- Chao Dong, "Image Super-Resolution Using Deep Convolutional Networks", IEEE Transactions on Pattern Analysis and Machine Intelligence, 38(2), pp.295~307 (2016), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Bulat, Adrian, "To learn image super-resolution, use a gan to learn how to do image degradation first", Proceedings of the European conference on computer vision (ECCV), pp.185~200 (2018)
- Kun Zhou, "Data-Parallel Octrees for Surface Reconstruction", IEEE Transactions on Visualization and Computer Graphics, 17(5), pp.669~681 (2011), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Gernot Riegler, "OctNet:Learning Deep 3D Representations at High Resolutions", IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2017)
- Peng-Shuai Wang, "O-CNN", ACM Transactions on Graphics, 36(4), pp.1~11 (2017), Association for Computing Machinery (ACM)[DOI]