当前位置: X-MOL 学术Mach. Vis. Appl. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Lightweight image super-resolution network using involution
Machine Vision and Applications ( IF 3.3 ) Pub Date : 2022-07-16 , DOI: 10.1007/s00138-022-01307-9
Jiu Liang, Yu Zhang, Jiangbo Xue, Yu Zhang, Yanda Hu

Recently, the single image super-resolution methods with deep and complex convolutional neural network structures have achieved remarkable performance. However, those approaches improve the performance at the cost of higher memory occupation, which are difficult to be applied for some resource-constrained devices. With the goal of minimizing parameters, an effective and efficient operator named involution is introduced in our proposed model, delivering enhanced performance at reduced cost compared to convolution-based counterparts. On the basis of involution, we propose two building blocks named RMFDB(Residual Mixed Feature Distillation Block) and CICB(Conv-Invo-Conv Block) for the main module and the reconstruction module respectively. RMFDB has the similar structure as the RFDB but with our involution layers. This block is much more lightweight and efficient than conventional convolution-based blocks. CICB combines the nearest-neighbor upsampling, convolution and involution layers. The final reconstruction quality is improved with little parameter cost. Experimental results demonstrate the effectiveness of the proposed model against the state-of-the-art (SOTA) SR methods. Our final model could achieve similar performance as the lightweight networks RFDN and PAN, but with only 224K parameters and 64.2G Multi-Adds with the scale factor of 2. The effectiveness of each proposed components is also validated by ablation study.



中文翻译:

使用对合的轻量级图像超分辨率网络

最近,具有深度和复​​杂卷积神经网络结构的单幅图像超分辨率方法取得了显着的性能。然而,这些方法以更高的内存占用为代价来提高性能,这很难应用于一些资源受限的设备。为了最小化参数,在我们提出的模型中引入了一个有效且高效的算子对合,与基于卷积的对应物相比,以更低的成本提供了增强的性能。在对合的基础上,我们分别为主模块和重构模块提出了两个构建模块RMFDB(Residual Mixed Feature Distillation Block)和CICB(Conv-Invo-Conv Block)。RMFDB 具有与 RFDB 类似的结构,但具有我们的对合层。这个块比传统的基于卷积的块更轻量级和更高效。CICB 结合了最近邻上采样、卷积和对合层。以很少的参数成本提高了最终的重建质量。实验结果证明了所提出的模型对最先进的(SOTA)SR方法的有效性。我们的最终模型可以实现与轻量级网络 RFDN 和 PAN 类似的性能,但只有 224K 参数和 64.2G Multi-Adds,比例因子为 2。每个提出的组件的有效性也通过消融研究得到验证。实验结果证明了所提出的模型对最先进的(SOTA)SR方法的有效性。我们的最终模型可以实现与轻量级网络 RFDN 和 PAN 类似的性能,但只有 224K 参数和 64.2G Multi-Adds,比例因子为 2。每个提出的组件的有效性也通过消融研究得到验证。实验结果证明了所提出的模型对最先进的(SOTA)SR方法的有效性。我们的最终模型可以实现与轻量级网络 RFDN 和 PAN 类似的性能,但只有 224K 参数和 64.2G Multi-Adds,比例因子为 2。每个提出的组件的有效性也通过消融研究得到验证。

更新日期:2022-07-17
down
wechat
bug