Lightweight Single Image Super-Resolution by Channel Split Residual Convolution
1
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
In recent years, deep convolutional neural networks have made significant progress in the research of singleimage super-resolution. However, it is difficult to be applied in practical computing terminals or embeddeddevices due to a large number of parameters and computational effort. To balance these problems, we proposeCSRNet, a lightweight neural network based on channel split residual learning structure, to reconstruct highresolutionimages from low-resolution images. Lightweight refers to designing a neural network with fewerparameters and a simplified structure for lower memory consumption and faster inference speed. At the sametime, it is ensured that the performance of recovering high-resolution images is not degraded. In CSRNet, wereduce the parameters and computation by channel split residual learning. Simultaneously, we propose adouble-upsampling network structure to improve the performance of the lightweight super-resolution networkand make it easy to train. Finally, we propose a new evaluation metric for the lightweight approaches named100_FPS. Experiments show that our proposed CSRNet not only speeds up the inference of the neural networkand reduces memory consumption, but also performs well on single image super-resolution.
Abstract
In recent years, deep convolutional neural networks have made significant progress in the research of singleimage super-resolution. However, it is difficult to be applied in practical computing terminals or embeddeddevices due to a large number of parameters and computational effort. To balance these problems, we proposeCSRNet, a lightweight neural network based on channel split residual learning structure, to reconstruct highresolutionimages from low-resolution images. Lightweight refers to designing a neural network with fewerparameters and a simplified structure for lower memory consumption and faster inference speed. At the sametime, it is ensured that the performance of recovering high-resolution images is not degraded. In CSRNet, wereduce the parameters and computation by channel split residual learning. Simultaneously, we propose adouble-upsampling network structure to improve the performance of the lightweight super-resolution networkand make it easy to train. Finally, we propose a new evaluation metric for the lightweight approaches named100_FPS. Experiments show that our proposed CSRNet not only speeds up the inference of the neural networkand reduces memory consumption, but also performs well on single image super-resolution.
참고문헌 (32)
- C. Dong, "Computer Vision – ECCV 2014", pp.184~199 (2014), Springer
- W. Shi, "Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.1874~1883 (2016)
- C. Dong, "Computer Vision – ECCV 2016", pp.391~407 (2016), Springer
- J. Kim, "Deeply-recursive convolutional network for image super-resolution", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.1637~1645 (2016)
- Y. Tai, "Image super-resolution via deep recursive residual network", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.2709~2798 (2017)
- N. Ahn, "Fast, accurate, and lightweight super-resolution with cascading residual network", Proceedings of the European Conference on Computer Vision (ECCV), pp.256~272 (2018)
- Z. Hui, "Fast and accurate single image super-resolution via information distillation network", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.723~731 (2018)
- "Multi-objective reinforced evolution in mobile neural architecture search" (2019)[URL]
- "Fast, accurate and lightweight super-resolution with neural architecture search" (2019)[URL]
- B. Zoph, "Learning transferable architectures for scalable image recognition", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.8697~8710 (2018)
- K. He, "Deep residual learning for image recognition", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.770~778 (2016)
- C. Ledig, "Photo-realistic single image super-resolution using a generative adversarial network", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.105~114 (2017)
- B. Lim, "Enhanced deep residual networks for single image superresolution", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp.1132~1140 (2017)
- Y. Zhang, "Image super-resolution using very deep residual channel attention networks", Proceedings of the European Conference on Computer Vision (ECCV), pp.294~310 (2018)
- S. Nah, "Deep multi-scale convolutional neural network for dynamic scene deblurring", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.257~265 (2017)
- "Mobilenets: efficient convolutional neural networks for mobile vision applications" (2017)[URL]
- B. Lim, "Enhanced deep residual networks for single image superresolution", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp.1132~1140 (2017)
- Z. Li, "Feedback network for image super-resolution", Proceedings of the IEEE, 2019, 3867~3876
- Z. Li, "Feedback network for image super-resolution", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.3867~3876 (2019)
- Z. Wang, "Image quality assessment : from error visibility to structural similarity", IEEE Transactions on Image Processing, 13(4), pp.600~612 (2004)
- X. Zhang, "Shufflenet : an extremely efficient convolutional neural network for mobile devices", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.6848~6856 (2018)
- K. Han, "Ghostnet : more features from cheap operations", Proceedings of the IEEE, 2020, 1577~1586
- E. Agustsson, "Ntire 2017 challenge on single image super-resolution : dataset and study", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, pp.1122~1131 (2017)
- M. Bevilacqua, "Low-complexity single-image superresolution based on nonnegative neighbor embedding", Proceedings of the British Machine Vision Conference (BMVC), pp.1~10 (2012)
- R. Zeyde, "Curves and Surfaces", pp.711~730 (2010), Springer
- D. Martin, "A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics", Proceedings 8th IEEE International Conference on Computer Vision (ICCV), pp.416~423 (2001)
- J. B. Huang, "Single image super-resolution from transformed self-exemplars", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.5197~5206 (2015)
- "Adam: a method for stochastic optimization" (2014)[URL]
- N. Ahn, "Fast, accurate, and lightweight super-resolution with cascading residual network", Proceedings of the European Conference on Computer Vision (ECCV), pp.256~272 (2018)
- Y. Zhang, "Residual dense network for image super-resolution", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.2472~2481 (2018)
- R. Lan, "MADNet : a fast and lightweight network for singleimage super resolution", IEEE Transactions on Cybernetics, 51(3), pp.1443~1453 (2020)
- L. Zhang, "Adaptive importance learning for improving lightweight image super-resolution network", International Journal of Computer Vision, 128(2), pp.479~499 (2020)