A Novel Fusion Pruning-Processed Lightweight CNN for Local Object Recognition on Resource-Constrained Devices

In recent years, smartphones and smart educational products have become important catalysts for the development of consumer electronics. However, deploying large convolutional neural networks (CNNs) models on resource-constrained devices like smartphones is impractical due to the lack of high-performance central processing units (CPUs), graphics processing units (GPUs), and large-capacity memory. To address these challenges, we propose a new Fusion Pruning (FP) method aimed at compressing and accelerating large CNN models by leveraging L2 norm and equivalent transformation of the receptive field. To validate the feasibility of the proposed method in language education, we have developed an application deployed on smartphones. This application enables offline object recognition without the need for additional hardware platforms like Jetson Xavier NX. Additionally, we have also explored the performance of the FP method in the field of household robotics, obtaining satisfactory results. Experimental results demonstrate that the proposed method can accomplish tasks on resource-constrained devices.

I. Introduction

Convolutional neural networks (CNNs) that are deployed on smartphones unlock a variety of novel applications empowered by deep learning, including object recognition-based second language learning [1]. With the rapid development of smartphones and broadband network technologies, the acquisition of multimedia information has become increasingly available [2], [3], resulting in a greater diversification of intelligent education formats. Among these formats, images represent the most intuitive and easily accessible learning materials [4], [5]. Intelligent recognition of image information by resource-constrained devices can be convenient for use in smart education systems. Directing learners to employ object recognition in real-world scenarios via smartphones, which then transforms these results into textual information and corresponding audio, stimulates greater interest in language learning and also enhances learning efficiency compared to traditional methods.