Keywords
MobileNetV2
Image Classification
Channel Attention
Edge Computing
Computational Optimization
Abstract
With the rapid development of edge computing and mobile intelligent devices, traditional deep convolutional neural networks (CNNs) suffer from excessive parameter volume, high computational complexity, and poor deployment adaptability on resource-constrained devices. To solve this problem, this paper proposes an improved lightweight neural network model based on MobileNetV2, which optimizes the inverted residual structure and introduces a dynamic channel attention mechanism. The model reduces redundant feature extraction operations while retaining effective image feature information, and balances classification accuracy and computational efficiency. A series of comparative experiments are conducted on the public CIFAR-10 dataset. The experimental results show that the improved model achieves a classification accuracy of 94.21%, with only 2.13M parameters and 0.58G floating-point operations (FLOPs). Compared with the original MobileNetV2, ResNet18 and other classic models, the proposed model has significant advantages in parameter quantity, computational consumption and inference speed, and is more suitable for real-time image classification tasks on edge devices. This study provides an effective technical reference for the lightweight deployment of computer vision algorithms.
References
[1] Que, H., Gao, H., Shan, W., Liu, M., An, J., Deng, F., ... & Mu, L. (2026). FM-SAM: individual tree crown delineation and classification based on Segmentation Anything Model (SAM) and YOLOv10 in UAV imagery for forest monitoring. Computers and Electronics in Agriculture, 240, 111162.
[2] Zhu, H., Huang, X., Gao, H., Jiang, M., Que, H., & Mu, L. (2025, July). A Wireless Collaborated Inference Acceleration Framework for Plant Disease Recognition. In International Conference on Intelligent Computing (pp. 331-341). Singapore: Springer Nature Singapore.
[3] Liu, M., Que, H., Yang, X., Zhang, K., Yu, Q., Yan, L., ... & Zhou, N. (2023). A Selective Bit Dropping and Encoding Co-Strategy in Image Processing for Low-Power Design in DRAM and SRAM. IEEE Journal on Emerging and Selected Topics in Circuits and Systems, 13(1), 48-57.
[4] Gao, H., Que, H., Au, H., Shan, W., Liu, M., Qin, Y., ... & Qiao, F. (2025). SenseExpo: Efficient Autonomous Exploration with Prediction Information from Lightweight Neural Networks. arXiv preprint arXiv:2503.16000.
[5] Wu, H., Liu, J., Zha, Z. J., Chen, Z., & Sun, X. (2019, August). Mutually Reinforced Spatio-Temporal Convolutional Tube for Human Action Recognition. In IJCAI (pp. 968-974).
[6] Li, W. (2026, August). Adaptive Prototypical Contrastive Learning for Time Series Clustering. In Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2 (pp. 2730-2741).
[7] Han, Z., Chen, W., Han, Y., Mao, R., & Qin, J. (2026). Fast Diversified Top-k Rule Discovery via User-Guided Embeddings. IEEE Transactions on Knowledge and Data Engineering.
[8] Zhang C, Gu X, Bhandari U, Lei J, Guo S, Kourouma M, Karoui A and Yang S (2022) Machine learning prediction of the mechanical properties of refractory multicomponent alloys based on a dataset of phase and first principles simulation. Front. Met. Alloy 1:1036656. doi: 10.3389/ftmal.2022.1036656
[9] Gu, X., & Wang, Q. (2025, July). Pathfence: Reducing cross-path dependencies in microservices. In Proceedings of the 34th International Symposium on High-Performance Parallel and Distributed Computing (pp. 1-13).
[10] Yao, T., Lu, B., Kowarschik, M., Yuan, Y., Zhao, H., Ourselin, S., ... & Qi, P. (2025). Advancing embodied intelligence in robotic-assisted endovascular procedures: A systematic review of ai solutions. IEEE Reviews in Biomedical Engineering.
[11] Yao, T., Lu, B., Kowarschik, M., Yuan, Y., Zhao, H., Ourselin, S., ... & Qi, P. (2025). Advancing embodied intelligence in robotic-assisted endovascular procedures: A systematic review of ai solutions. IEEE Reviews in Biomedical Engineering.
[12] Chaoyong Ma, Nan Si, Xuewei Song, Chen Liang and Haoming Luo, A Bearing Fault Diagnosis Method Based on Concise Empirical Wavelet Transform, International Journal of Comprehensive Engineering，2023, 12(1), 1-6
[13] Yuyan Li, Lin Chen, Mengyu Yang, Qiaomei Li and Taifu Li, Polarization Converters Composed of Metasurfaces with Independent Manipulation of Polarization and Phase, International Journal of Comprehensive Engineering，2023, 12(1), 7-14
[14] Jiaxun Du, Zhenbao Fu, Jiachen Liu, Huaqing Wang, Zhentao Ge and Liuyang Song, Bearing Fault Diagnosis Method Based on Attention Residual Network Under Small Sample Condition, International Journal of Comprehensive Engineering，2023, 12(1), 15-24
[15] Jiawei Lin, Changkun Han, Liuyang Song, Wei Lu and Huaqing Wang, Time Domain Feature Enhancement Model Based on Correlated B-spline Wavelet Base Convolutional Sparse, International Journal of Comprehensive Engineering，2023, 12(1), 25-33
[16] Bing Wang, Haihong Tang, Xiaojia Zu and Wuwei Feng, Application of Parameter Adaptive VMD in Bearing Fault Diagnosis, International Journal of Comprehensive Engineering，2023, 12(1), 34-42
[17] Xiaojia Zu, Haihong Tang and Bing Wang, Research on Intelligent Diagnosis Based on Support Vector Machine for Structural Fault, International Journal of Comprehensive Engineering，2024, 13(1), 1-12.
[18] Xi Qiao, Di Zhao, Nan Si, Jifeng Sui and Yonggang Xu, Weighting Spectrum Editing Based on Amplitude Mapping and Its Application in Composite Fault Diagnosis of Rolling Bearings, International Journal of Comprehensive Engineering，2024, 13(1), 13-21
[19] Xixin Chen, Ming Tu, Lei Su and Ke Li, Rolling Bearing Fault Diagnosis Based on Digital Twin Method, International Journal of Comprehensive Engineering，2024, 13(1), 22-39
[20] Jianheng Liang, Yang Yang, Ran Zhang, Guican Wu, Yuanyuan Yang, TaiFu Li, Honglin Duan and Yuyan Li, Research on Perfume User Experience Evaluation Based on Human Machine Hybrid Generation Adversarial Network, International Journal of Comprehensive Engineering，2024, 13(1), 40-54
[21] Han Feng, Xia Qin and Hongtao Xue, A Novel BN and DST Fusion Framework for Fault Identification of In-wheel Motor, International Journal of Comprehensive Engineering，2024, 13(1), 55-62
[22] Feng Sun, Zhenguo Song, Dinghao Dong, Shuai Shen, Jie Cai, Heng Luo and Jianhui Zhou, A Variational Time-domain Decomposition Method for Vibration Signals Based on Exponential Adaptive Weighted Moving Average Optimization, International Journal of Comprehensive Engineering，2025, 14(1), 1-12
[23] Zhichao Qiu, Zhijia Yan, Jinhua Huang, Yewen Chen, Yaotiao Ren, Shimin Yu and Xuewei Song,Bearing Fault Diagnosis Based on Adaptive Multiple Synchronous Compressed Transform and Twin Network, International Journal of Comprehensive Engineering，2025, 14(1), 13-22
[24] Xuefei Ming, Zhenyu Pan, Gang Wang, Yong Ji, Lei Su and Ke Li, Sparse Wavelet Neural Network with Frequency-Domain Gating for Ultrasonic Vibration-Based Flip-Chip Defect Detection, International Journal of Comprehensive Engineering，2025, 14(1), 23-39
[25] Minggang Tong, Xin Wang, Wenhua Zhang, Qi Hu and Zengwei Lyu, Intelligent Fault Diagnosis of Dynamic Equipment in Public Auxiliary Systems Based on Multimodal Deep Q-Networks, International Journal of Comprehensive Engineering，2025, 14(1), 40-54
[26] Lijun Yan, Chaoyong Ma, Heng Zhang, Mengjie Xie and Rongbin Zhang, Multi-objective Optimization of Solenoid Valves Integrating AHP Decision-making and Orthogonal Design, International Journal of Comprehensive Engineering，2026, 15(1), 1-12
[27] Jiaqi Guan, Jiachen Liu, Huaqing Wang, Liuyang Song, Hassan Sajjad and Zhilong Gao, Research on Rolling Bearing Remaining Useful Life Prediction Method Based on Knowledge Distillationnternational, Journal of Comprehensive Engineering，2026, 15(1), 13-23
[28] Tian Teng and Xiao Zhang, Design Principles and Feasibility Assessment of an Intelligent Spare-Parts Shelving System for Shipboard Storage Environments, Journal of Comprehensive Engineering，2026, 15(1), 24-30
[29] Guo, Z., Fu, J., Bhanushali, S., Zeng, Z., Banerjee, I., & Sanyal, A. (2026, May). Balancing Speed and Accuracy for Robust Analog-Mixed Signal Circuit Design using Closed-Loop Reinforcement Learning with Ensemble Neural Network Surrogates. In 2026 IEEE International Symposium on Circuits and Systems (ISCAS) (pp. 803-807). IEEE.
[30] Quan, Q., Gao, Y., & Wang, Q. (2025). The impact of teacher emotional support on students' engagement in AI-mediated English learning environments: The mediating role of resilience and self-efficacy. Acta Psychologica, 260, 105766.
[31] Wang, Z., Yu, J., Liu, H., Zheng, Z., Jin, Y., Li, S., ... & Zhang, K. (2025, July). Generative Music Models’ Alignment with Professional and Amateur Users’ Expectations. In Findings of the Association for Computational Linguistics: ACL 2025 (pp. 6909-6920).
[32] Xu, Y., Cui, J., Zhu, Z., Shang, H., Luan, S., Xu, M., ... & Zhu, S. (2026). Wam-flow: Parallel coarse-to-fine motion planning via discrete flow matching for autonomous driving. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 24918-24928).
[33] Zheng, X., Hu, S., Dwyer, V., Barrett, L., & Derakhshani, M. (2026). Joint attention mechanism learning to facilitate opto-physiological monitoring during physical activity. Biomedical Signal Processing and Control, 113, 108949.
[34] Dong, Guangyuan, Hong, Ziwei, and Liu, Shenghao, et al. Dynamic Interaction-Aware and Causality-Disentangled Framework for Multimodal Sentiment Analysis. arXiv preprint arXiv:2605.30994, 2026.
[35] Su, J., He, C., Zhu, F., Xu, X., Guan, D., & Si, C. (2024, September). Hpff: Hierarchical locally supervised learning with patch feature fusion. In European Conference on Computer Vision (pp. 293-309). Cham: Springer Nature Switzerland.