Quantizing Oriented Object Detection Network via Outlier-Aware Quantization and IoU Approximation
Author(s): Zhao, MX (Zhao, Mingxin); Ning, K (Ning, Ke); Yu, SM (Yu, Shuangming); Liu, LY (Liu, Liyuan); Wu, NJ (Wu, Nanjian)
Source: IEEE SIGNAL PROCESSING LETTERS Volume: 27 Pages: 1914-1918 DOI: 10.1109/LSP.2020.3031490 Published: 2020
Abstract: In recent years, a large number of quantization schemes have been proposed for compressing convolutional neural networks (CNN). However, most of them have the following problems: 1) when there are outliers in the weight, post-training quantization cannot obtain the ideal effect, and the accuracy loss is unavoidable; 2) quantizing the non-maximum suppression (NMS) stage of oriented object detection networks is non-trivial so that such networks are difficult to deploy on edge computing devices that only support integer operations. In this letter, we propose the outlier-aware quantization (OAQ) to boost the robustness of the post-training quantization method. Besides, we design a multilayer perceptron network to approximate the intersection-over-union (IoU) of rotated boxes, making the NMS stage can be deployed on integer-arithmetic-only devices. The experiment results demonstrate that our solution outperforms the widely used post-training quantizationmethod. Meanwhile, to the best of our knowledge, this is the first study that focuses on the optimization and quantization of the NMS stage of oriented object detection networks.
Accession Number: WOS:000587912400014
ISSN: 1070-9908
eISSN: 1558-2361
Full Text: https://ieeexplore.ieee.org/document/9226069