基于迁移学习的蔬菜图像识别方法
PDF下载 (447)赖佩霞,王晓东,章联军.基于迁移学习的蔬菜图像识别方法[J].宁波大学学报(理工版),2019,32(5):36-41.DOI:
LAI Peixia,WANG Xiaodong,ZHANG Lianjun.Vegetable image recognition based on transfer learning[J].Journal of Ningbo University(Natural Science & Engineering Edition),2019,32(5):36-41.DOI:
| Title: | Vegetable image recognition based on transfer learning |
| 作者: | 赖佩霞, 王晓东, 章联军 |
| Author(s): | LAI Peixia, WANG Xiaodong, ZHANG Lianjun |
| 关键词: | 蔬菜图像识别; 卷积神经网络; 迁移学习; 小样本 |
| Keywords: | vegetable image recognition; convolution neural network; transfer learning; small samples |
| 分类号: | TP391.4 |
| 文献标识码: | A |
| 摘要: | 为解决蔬菜识别领域缺少带标签样本的问题, 提出了一种基于迁移学习的图像识别方法. 首先, 将原始数据集利用数据增强扩大样本数据量后引入到大规模数据集上的预训练模型. 针对迁移过程中高层特征的领域特定性导致的网络泛化性能差, 通过加入两层自适应层参数初始化后重新训练得到基本模型; 对该基本模型再利用参数冻结的迁移方式进一步调优参数, 得到用于蔬菜图像识别的最终网络模型. 实验表明, 基于CaffeNet和ResNet10两个小型网络的迁移策略可以较好地处理小样本的蔬菜图像识别, 训练得到的模型准确率分别为94.97％、96.69％. 与其他迁移算法及传统的神经网络方法相比, 该算法具有更高的识别性以及更强的鲁棒性. |
| Abstract: | To solve the problem of lacking the labeled samples in vegetable recognition domain, a method for image recognition based transfer learning is proposed. Firstly, raw data are expanded by the data augmentation technique, then pre-trained models on the large-scale data sets are introduced to the target data set to train for the base model, where initializing two additional adaptive layers is performed. This step is aimed to solve the poor generalization performance caused by the transferred domain specific high-level features. Next, the final model is obtained for recognizing vegetable images by adopting the fine-tuning method with several layers frozen. Experimental results show that, based on CaffeNet and ResNet10, the proposed transfer approach reaches the accuracy of 94.97％, 96.69％, respectively, and can effectively process small samples in vegetable image recognition with higher accuracy and better robustness comparing to other transferring algorithms and the conventional convolution neural networks. |
| 参考文献 /References: | [1] Gu H, Wang D. A content-aware fridge based on RFID in smart home for home-healthcare[EB/OL]. [2018-08-28]. https://ieeexplore.ieee.org/document/4809580. [2] Sharma A, Paliwal K K. Linear discriminant analysis for the small sample size problem: An overview[J]. International Journal of Machine Learning & Cybernetics, 2014, 6(3):443-454. [3] Bolle R M, Connell J H, Haas N, et al. Veggievision: A produce recognition system[EB/OL]. [2018-08-28]. https:// ieeexplore.ieee.org/document/572062?arnumber=572062&tag=1. [4] 庄福振, 罗平, 何清, 等. 迁移学习研究进展[J]. 软件学报, 2015, 26(1):26-39. [5] Yosinski J, Clune J, Bengio Y, et al. How transferable are features in deep neural networks?[EB/OL]. [2018-08-28]. https://www.researchgate.net/publication/268079628_How_ transferable_are_features_in_deep_neural_networks. [6] 王柯力, 袁红春. 基于迁移学习的水产动物图像识别方法[J]. 计算机应用, 2018, 38(5):1304-1308. [7] 段萌, 王功鹏, 牛常勇. 基于卷积神经网络的小样本图像识别方法[J]. 计算机工程与设计, 2018, 39(1):224- 229. [8] Hou S, Feng Y, Wang Z. VegFru: A domain-specific dataset for fine-grained visual categorization[EB/OL]. [2018-08-28]. https://ieeexplore.ieee.org/document/8237328. [9] 曾维亮, 林志贤, 陈永洒. 基于卷积神经网络的智能冰箱果蔬图像识别的研究[J]. 微型机与应用, 2017, 36(8): 56-59. [10] Shie C K, Chuang C H, Chou C N, et al. Transfer representation learning for medical image analysis[EB/ OL]. [2018-08-28]. https://www.researchgate.net/publication/ 308298066_Transfer_representation_learning_for_medical_image_analys-is. [11] Xie M, Jean N, Burke M, et al. Transfer learning from deep features for remote sensing and poverty mapping[EB/OL]. [2018-08-28]. https://dlacm.org/citation. cfm?id=3016457. [12] Dai W, Yang Q, Xue G R, et al. Boosting for transfer learning[EB/OL]. [2018-08-28]. http://citeseerx.ist.psu. edu/ viewdoc/summary?doi=10.1.1.76.7577. [13] Ando R K, Zhang T. A high-performance semi-supervised learning method for text chunking[EB/OL]. [2018-08-28]. https://dl.acm.org/citation.cfm?id=1219841. [14] Mohapatra R K. Majhirning in feedforward neural network[J]. International Journal of Multimedia & Ubiquitous Engineering, 2014, 9(8):149-156. [15] Jia Y, Shelhamer E, Donahue J, et al. Caffe: Convolutional architecture for fast feature embedding [EB/OL]. [2018-08-28]. https://arxiv.org/abs/1408.5093. |
| 备注/Memo: | 收稿日期: 2018－12－11. 宁波大学学报（理工版）网址: http://journallg.nbu.edu.cn/ 基金项目: 国家科技支撑计划项目（2012BAH67F01）; 国家自然科学基金（U1301257）; 浙江省自然科学基金（LY17F010005）. 宁波大学学报（理工版）网址：http://journallg.nbu.edu.cn/第一作者: 赖佩霞（1993－）, 女, 浙江宁波人, 在读硕士研究生, 主要研究方向: 图像模式识别与深度学习. E-mail: jula199@163.com *通信作者: 王晓东（1970－）, 男, 浙江上虞人, 教授, 主要研究方向: 网络通信及计算机软件. E-mail: wxd@nbu.edu.cn |
