Basic Binary Convolution Unit for Binarized Image Restoration Network

TL;DR

This paper introduces a Basic Binary Convolution Unit (BBCU) tailored for binarized image restoration networks, systematically analyzing components and designing variants to improve efficiency and performance on resource-limited devices.

Contribution

The study provides a systematic analysis of binary convolution components for IR, and proposes a novel BBCU with variants optimized for different IR network parts, outperforming existing BNNs.

Findings

BBCU significantly outperforms other BNNs in IR tasks

Residual connections reduce information loss in binary convolution

Proper placement of activation functions greatly impacts BNN performance

Abstract

Lighter and faster image restoration (IR) models are crucial for the deployment on resource-limited devices. Binary neural network (BNN), one of the most promising model compression methods, can dramatically reduce the computations and parameters of full-precision convolutional neural networks (CNN). However, there are different properties between BNN and full-precision CNN, and we can hardly use the experience of designing CNN to develop BNN. In this study, we reconsider components in binary convolution, such as residual connection, BatchNorm, activation function, and structure, for IR tasks. We conduct systematic analyses to explain each component's role in binary convolution and discuss the pitfalls. Specifically, we find that residual connection can reduce the information loss caused by binarization; BatchNorm can solve the value range gap between residual connection and binary convolution; The position of the activation function dramatically affects the performance of BNN. Based on our findings and analyses, we design a simple yet efficient basic binary convolution unit (BBCU). Furthermore, we divide IR networks into four parts and specially design variants of BBCU for each part to explore the benefit of binarizing these parts. We conduct experiments on different IR tasks, and our BBCU significantly outperforms other BNNs and lightweight models, which shows that BBCU can serve as a basic unit for binarized IR networks. All codes and models will be released.