Architecture-aware Network Pruning for Vision Quality Applications
arXiv:1908.02125
The paper introduces an iterative, architecture-aware pruning method that adaptively thresholds weights while monitoring image quality, achieving significant reductions in computation and memory bandwidth for low-light imaging and super-resolution tasks without degrading visual quality.
Abstract
Convolutional neural network (CNN) delivers impressive achievements in computer vision and machine learning field. However, CNN incurs high computational complexity, especially for vision quality applications because of large image resolution. In this paper, we propose an iterative architecture-aware pruning algorithm with adaptive magnitude threshold while cooperating with quality-metric measurement simultaneously. We show the performance improvement applied on vision quality applications and provide comprehensive analysis with flexible pruning configuration. With the proposed method, the Multiply-Accumulate (MAC) of state-of-the-art low-light imaging (SID) and super-resolution (EDSR) are reduced by 58% and 37% without quality drop, respectively. The memory bandwidth (BW) requirements of convolutional layer can be also reduced by 20% to 40%.
Accepted to be Published in the 26th IEEE International Conference on Image Processing (ICIP 2019). Updated to contain the IEEE copyright notice