基于域泛化的轻量化图像分类算法

doi:10.19678/j.issn.1000-3428.0068403

摘要/Abstract

摘要： 针对现有睡岗数据集少，现阶段分类算法泛化性差，推理速度慢等问题，构建了一份包含4708张图像的睡岗数据集，用于验证模型识别精度和泛化能力，并提出了一种基于域泛化的轻量化图像分类算法Stable_MobileNet。首先，对输入的图片填充短边，使其保持图像中的人物比例，然后再进行图像增强和随机擦除，用于扩充数据集；其次，引入高效的ECA注意力模块改进MobileNetv3_large网络；最后，使用稳定学习StableNet方法提高模型的泛化性，通过学习训练样本的权重来消除特征之间的依赖关系，这有助于模型摆脱环境的变化，更专注于人物特征。在睡岗数据集上的实验结果表明，Stable_MobileNet平均推理速度更快，识别精度可达93.56%，比MobileNetv3_large提高了2.23%；在样本分布异于训练样本分布的测试集中，该算法的识别精度提高了2.23%。

Abstract: Aiming at the problems of few Sleeping on Duty dataset, poor generalization of current classification algo-rithms, and slow inference speed. A Sleeping on Duty dataset containing 4708 images is constructed to verify the recognition accuracy and generalization ability of the model, and a lightweight image classification algo-rithm based on domain generalization, Stable_MobileNet, is proposed. Firstly, the input images are filled with short edges to keep the proportion of people in the images, and then image enhancement and random erasure are performed for expanding the dataset; Secondly, the introduction of efficient ECA module to improve the MobileNetv3_large network; Finally, the Stable learning StableNet method is used to improve the generali-zation of the model by learning the weights of the training samples to eliminate the dependency between features, which helps the model get rid of the environment and more focused on the character features. Ex-perimental results on the Sleeping on Duty dataset show that Stable_MobileNet has faster inference on average, and the recognition accuracy can reach 93.56%, which is 2.23% better than MobileNetv3_large; In the test set where the sample distribution is different from the training sample distribution, the recognition accuracy of the algorithm is improved by 2.23%.

张倡倡, 吕卫东, 蔡子杰 , 刘炎奎. 基于域泛化的轻量化图像分类算法[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0068403.

ZHANG Chang-chang, LV Wei-dong, CAI Zi-jie, LIU Yan-kui. Lightweight Image Classification Algorithm Based on Domain Generalization[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0068403.

参考文献

[1] 王佟, 韩效忠, 邓军, 等. 论中国煤炭地质勘查工作在新条件下的定位与重大研究问题[J]. 煤田地质与勘探, 2023, 51(2): 1−19. Wang T, Han X Z, Deng J, et al. Orientation and major research problems of coal geological exploration in China under new conditions[J]. Coal Geology & Exploration, 2023, 51(2): 1-19
[2] 程德强, 寇旗旗, 江鹤等. 全矿井智能视频分析关键技术综述[J]. 工矿自动化, 2023, 49(11): 1-21. Cheng D Q, Kou Q Q, Jiang H, et al. Overview of key technologies for mine-wide intelligent video analysis[J]. Journal of Mine Automation, 2023, 49(11): 1-21.
[3] 张书伟. 面向智能安防场景的人体行为识别算法研究及应用[D]. 西安电子科技大学, 2021. Zhang S W. Research and Application of Human Body Behavior Recognition Algorithm for Intelligent Security Scenes[D]. XIDIAN UNIVERSITY, 2021.
[4] He K, Zhang X, Ren S, et al. Delving deep into rectifiers: Surpassing human-level performance on imagenet classification[C]//Proceedings of the IEEE international conference on computer vision. IEEE, 2015: 1026-1034.
[5] Sharma S, Guleria K, Deep learning models for image classification: comparison and applications[C]//2022 2nd International Conference on Advance Computing and Innovative Technologies in Engineering (ICACITE). IEEE, 2022: 1733-1738.
[6] 刘雪 , 沈长盈 , 吕学泽 , 等 . 基于改进 MobileNetV3-Large 的鸡蛋新鲜度识别模型[J]. 农业工程学报, 2022, 38(17): 196-204. Liu X, Shen C Y, Lv X Z, et al. Recognizing egg freshness using an improved MobileNetV3-Large[J]. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(17): 196-204.
[7] 李小平, 白超. 基于深度学习的司机疲劳驾驶检测方法研究[J]. 铁道学报, 2021, 43(06): 78-87. Li X P, Bai C. Research on Driver Fatigue Driving Detection Method Based on Deep Learning[J]. Journal of the China Railway Society, 2021, 43(06): 78-87.
[8] Zhou Y, Chen S, Wang Y, et al. Review of research on lightweight convolutional neural networks[C]//2020 IEEE 5th Information Technology and Mechatronics Engineering Conference (ITOEC). IEEE, 2020: 1713-1720.
[9] Zhang X, Cui P, Xu R, et al. Deep Stable learning for Out-Of-Distribution generalization[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2021: 5372-5382.
[10] Li Y, Liu J, Wang L. Lightweight network research based on deep learning: A review[C]//2018 37th Chinese control conference (CCC). IEEE, 2018: 9021-9026.
[11] 毕鹏程, 罗健欣, 陈卫卫. 轻量化卷积神经网络技术研究[J]. 计算机工程与应用, 2019, 55(16): 25-35. Bi P C, Luo J X, Chen W W. i. Research on lightweight convolutional neural network technology. Computer Engineering and Applications, 2019, 55(16): 25-35.
[12] Cheng Y, Wang D, Zhou P, et al. A survey of model compression and acceleration for deep neural networks[J]. arXiv preprint arXiv:1710.09282, 2017.
[13] Howard A G, Zhu M, Chen B, et al. Mobilenets: efficient convolutional neural networks for mobile vision applications[J]. arXiv preprint arXiv:1704.04861, 2017.
[14] Sandler M, Howard A, Zhu M, et al. Mobilenetv2: inverted residuals and linear bottlenecks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2018: 4510-4520.
[15] Tan M, Chen B, Pang R, et al. Mnasnet: Platform-aware neural architecture search for mobile[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE, 2019: 2820-2828.
[16] Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. 2018: 7132-7141.
[17] Howard A, Sandler M, Chu G, et al. Searching for mobilenetv3[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. IEEE, 2019: 1314-1324.
[18] Wang J, Lan C, Liu C, et al. Generalizing to unseen domains: A survey on domain generalization[J]. IEEE Transactions on Knowledge and Data Engineering, 2022: 1-1.
[19] Zhou K, Liu Z, Qiao Y, et al. Domain generalization: A survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022.
[20] Carlucci F M, D'Innocente A, Bucci S, et al. Domain generalization by solving jigsaw puzzles[C]//Proceedings of the IEEE/CVF Conferenceon Computer Vision and Pattern Recognition. 2019: 2229-2238.
[21] Zhong Z, Zheng L, Kang G, et al. Random erasing data augmentation[C]//Proceedings of the AAAI Conference on Artificial Intelligence. AAAI, 2020, 34(07): 13001-13008.
[22] Wang Q, Wu B, Zhu P, et al. ECA-Net: efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2020: 11534-11542.
[23] Gretton A, Fukumizu K, Teo C, et al. A kernel statistical test of independence[J]. Advances in Neural Information Processing Systems, 2008, 20: 585-592.
[24] Strobl E V, Zhang K, Visweswaran S. Approximate kernel-based conditional independence tests for fast non-parametric causal discovery[J]. Journal of Causal Inference, 2019, 7(1): 1-24.
[25] Li D, Yang Y, Song Y Z, et al. Deeper, broader and artier domain generalization[C]//Proceedings of the IEEE International Conference on Computer Vision. IEEE, 2017: 5542-5550.

选择文件类型/文献管理软件名称

选择包含的内容