首页 > 分享 > 基于改进ResNet50模型的大宗淡水鱼种类识别方法

基于改进ResNet50模型的大宗淡水鱼种类识别方法

萌宠菠菠乐园
2024-12-03 11:42

摘要: 针对传统鱼类识别方法存在特征提取复杂、算法可移植性差等不足，该研究提出了一种基于改进ResNet50模型的淡水鱼种类识别方法。研究以鳙鱼、鳊鱼、鲤鱼、鲫鱼、草鱼、白鲢6种大宗淡水鱼为对象，通过搭建淡水鱼图像采集系统获取具有单一背景的淡水鱼图像，同时通过互联网搜索具有干扰背景的淡水鱼图像，共同构建淡水鱼图像数据集；再对淡水鱼图像进行预处理，以增加样本多样性；构建改进ResNet50模型，增加全连接层Fc1以及Dropout，引入迁移学习机制训练模型，同时选择CELU作为激活函数提高神经网络表达能力，通过Adam优化算法更新梯度，并嵌入余弦退火方法衰减学习率。为验证改进ResNet50模型的准确率等性能，对6种淡水鱼进行种类识别，结果表明：在单次验证方法下，选用包含单一背景图像和干扰背景图像构成的淡水鱼图像数据集训练模型，识别准确率为96.94%，比经典模型提高1.22%，单张淡水鱼图像样本的平均检测时间为0.234 5 s；在四折交叉验证下，选用具有单一背景的图像数据集，模型的识别准确率为100%，选用包含单一背景图像和干扰背景图像的淡水鱼图像数据集，模型的识别准确率为96.20%，说明模型具有较好的泛化性能和鲁棒性。针对混淆矩阵的可视化结果表明：改进的ResNet50模型具有通用的结构和训练方式，对不同背景下的淡水鱼进行种类识别具有较高的准确率，可为淡水鱼种类识别提供技术借鉴。

Abstract: Abstract: Species identification of freshwater fish has a wide range of applications in most fields, such as breeding, fishing, and processing. However, most traditional algorithms of fish identification cannot meet the ever-increasingly high requirements in recent years, such as simple feature extraction, high accuracy, and portability. In this study, a new identification algorithm was proposed for the freshwater fish species using an improved ResNet50 model. Six types of freshwater fish were taken as the research objects, including the bighead, bream, carp, crucian, grass carp, and silver carp. An image acquisition system was established for the freshwater fish images with a single background. As such, an image dataset of freshwater fish was constructed to joint those images with interference background on the Internet. A Pytorch framework was then selected to perform image preprocessing of freshwater fish for the sample diversity. An improved ResNet50 model was thus built to identify the freshwater fish species. Firstly, the fully connected layer Fc1 and Dropout were added, while the migration learning was introduced to train the model. Secondly, CELU was selected as the activation function to improve the expression of the neural network. Finally, Adam optimization was used to update the gradient. A cosine annealing was also embedded to attenuate the learning rate. In addition, the hyperparameters of the model were optimized in the multiple model training. Correspondingly, six kinds of freshwater fish were identified to verify the accuracy and performance of the improved ResNet50 model. A single validation test under a four-fold cross-validation model was carried out to train and evaluate the model. The confusion matrix was used to visualize the recognition of each type of fish. The results showed that: the image dataset of freshwater fish consisting of a single and interference background images was selected to train the model under the single validation, where the accuracy rate was 96.94%, 1.22% higher than before. The average detection time was 0.2345s for a single freshwater fish image. The accuracy rate of the model was 100% under the four-fold cross-validation, when the dataset of the freshwater fish image was selected with a single background. By contrast, the accuracy rate of the model was 96.20%, when the dataset of freshwater fish image consisted of a single and interference background, indicating an excellent general performance and robustness. The accuracy, recall and F1 score of each type of freshwater fish were relatively high visualized to the confusion matrix, when the model was trained on the freshwater fish image and a single background dataset, indicating the superior performance of the model. The improved ResNet50 model presented a general structure and training, while a high accuracy rate under different backgrounds. The finding can provide a sound technical reference for the identification of freshwater fish species in intelligent aquaculture.

[1] 于秀娟，徐乐俊，吴反修. 中国渔业统计年鉴[M]. 北京：中国农业出版社，2020. [2] 张军文，陈庆余，欧阳杰，等. 中国淡水鱼前处理加工技术研究进展[J]. 安徽农业科学，2018，46(21)：25-28.Zhang Junwen, Chen Qingyu, Ouyang Jie, et al. Research progress in freshwater fish pretreatment and processing technology in China[J]. Anhui Agricultural Sciences, 2018, 46(21): 25-28. (in Chinese with English Abstract) [3] 邹伟. 淡水鱼连续式去头尾装置的设计与试验研究[D]. 武汉：华中农业大学，2018.Zou Wei. Design and Experimental Research of Continuous Head and Tail Device for Freshwater Fish[D]. Wuhan: Huazhong Agricultural University, 2018. (in Chinese with English Abstract) [4] 谭鹤群，高星星. 链式剖鱼机工艺参数优化试验[J]. 农业工程学报，2011，27(6)：332-336.Tan Hequn, Gao Xingxing. Optimization test of technological parameters of chain fish slitter[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2011, 27(6): 332 -336. (in Chinese with English Abstract) [5] 杨东海，张胜茂，汤先峰. 基于机器视觉技术的鱼类识别研究进展[J]. 渔业信息与战略，2019，34(2)：37-45.Yang Donghai, Zhang Shengmao, Tang Xianfeng. Research progress of fish recognition based on machine vision technology[J]. Fisheries Information and Strategy, 2019, 34(2): 37-45. (in Chinese with English Abstract) [6] 陈从平，吴杞，吴喆，等. 一种基于图像处理的鱼类自动分类方法：以四种淡水鱼为例[J]. 软件工程，2018，21(12)：7-11.Chen Congping, Wu Qi, Wu Zhe, et al. An automatic fish classification method based on image processing: A case study of four freshwater fish[J]. Software Engineering, 2018, 21(12): 7-11. (in Chinese with English Abstract) [7] 张志强，牛智有，赵思明. 基于机器视觉技术的淡水鱼品种识别[J]. 农业工程学报，2011，27(11)：388-392.Zhang Zhiqiang, Niu Zhiyou, Zhao Siming. Freshwater fish species identification based on machine vision technology[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2011, 27(11): 388-392. (in Chinese with English Abstract) [8] 万鹏，潘海兵，龙长江，等. 基于机器视觉技术淡水鱼品种在线识别装置设计[J]. 食品与机械，2012，28(6)：164-167.Wan Peng, Pan Haibing, Long Changjiang, et al. Design of online identification device for freshwater fish species based on machine vision technology[J]. Food and Machinery, 2012, 28(6): 164-167. (in Chinese with English Abstract) [9] 吴一全，殷骏，戴一冕，等. 基于蜂群优化多核支持向量机的淡水鱼种类识别[J]. 农业工程学报，2014，30(16)：312-319.Wu Yiquan, Yin Jun, Dai Yimian, et al. Freshwater fish species identification based on bee colony optimization multi-core support vector machine[J]. Transactions of the Chinese Society of Agricultural Engineering (Transactions of the CSAE), 2014, 30(16): 312-319. (in Chinese with English Abstract) [10] 胡灿. 基于ELM算法的鲜活淡水鱼图像自动归类研究[D]. 湘潭：湘潭大学，2017.Hu Can. Research on Automatic Classification of Fresh Freshwater Fish Images based on ELM Algorithm[D]. Xiangtan: Xiangtan University, 2017. (in Chinese with English Abstract) [11] 涂兵，王锦萍，王思成，等. 基于背部轮廓相关系数算法的淡水鱼种类识别研究[J]. 计算机工程与应用，2016，52(16)：162-166.Tu Bing, Wang Jinping, Wang Sicheng, et al. Identification of freshwater fish species based on back contour correlation coefficient algorithm[J]. Computer Engineering and Application, 2016, 52(16): 162-166. (in Chinese with English Abstract) [12] White D J, Svellingen C, Strachan N J C. Automated measurement of species and length of fish by computer vision[J]. Fisheries Research, 2006, 80(4): 203-210. [13] Larsen R, Olafsdottir H. Shape and texture based classification of fish species[C]. //Scandinavian Conference on Image Analysis. Springer-Verlag, 2009: 745-749. [14] Alsmadi M K, Omar K B, Noah S A, et al. Fish recognition based on robust features extraction from size and shape measurements using neural network[J]. Journal of Computer Science, 2010, 6(10): 1059-1065. [15] 陈文辉，蔡妹姝，严松，等. 基于深度卷积神经网络与迁移学习的鱼类分类识别[J]. 福建师大福清分校学报，2019(5)：30-38.Chen Wenhui, Cai Meishu, Yan Song, et al. Fish classification and recognition based on deep convolution neural network and transfer learning[J]. Journal of Fuqing Branch of Fujian Normal University, 2019(5): 30-38. (in Chinese with English Abstract) [16] 段青玲，李道亮，姚红，等. 基于图像处理的目标鱼识别方法及系统：201310317701. 3[P]. 2016-07-06. [17] 胡涛. 基于深度学习的鱼类识别研究[D]. 杭州：浙江工业大学，2019.Hu Tao. Research on Fish Recognition Based on Deep Learning[D]. Hangzhou: Zhejiang University of Technology, 2019. (in Chinese with English Abstract) [18] 顾郑平，朱敏. 基于深度学习的鱼类分类算法研究[J]. 计算机应用与软件，2018，35(1)：200-205.Gu Zhengping, Zhu Min. Research on fish classification algorithm based on deep learning[J]. Computer Applications and Software, 2018, 35(1): 200-205. (in Chinese with English Abstract) [19] 陈英义，龚川洋，刘烨琦，等. 基于FTVGG16卷积神经网络的鱼类识别方法[J]. 农业机械学报，2019，50(5)：223-231.Chen Yingyi, Gong Chuanyang, Liu Yeqi, et al. Fish recognition method based on FTVGG16 convolutional neural network[J]. Transactions of The Chinese Society for Agricultural Machinery(Transactions of the CSAM), 2019, 50(5): 223-231. (in Chinese with English Abstract) [20] 王文成，蒋慧，乔倩，等. 基于ResNet50网络的十种鱼类图像分类识别研究[J]. 农村经济与科技，2019，30(19)：60-62.Wang Wencheng, Jiang Hui, Qiao Qian, et al. Classification and recognition of ten fish images based on ResNet50 network[J]. Rural Economy and Technology, 2019, 30(19): 60-62. (in Chinese with English Abstract) [21] Vaneeda A, Olav H N, Shale R, et al. Fish species identification using a convolutional neural network trained on synthetic data[J]. ICES Journal of Marine Science, 2019, 76(1): 342-349. [22] Banan A, Nasiri A, Taheri-Garavand A. Deep learning-based appearance features extraction for automated carp species identification[J]. Aquacultural Engineering, 2020, 89: 134-139. [23] 胡越，罗东阳，花奎，等. 关于深度学习的综述与讨论[J]. 智能系统学报，2019，14(1)：1-19.Hu Yue, Luo Dongyang, Hua Kui, et al. Overview and discussion on deep learning[J]. Journal of Intelligent Systems, 2019, 14(1): 1-19. (in Chinese with English Abstract) [24] He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//IEEE Conference on Computer Vision & Pattern Recognition. IEEE Computer Society, 2016, 1-12. [25] Barron J T. Continuously differentiable exponential linear units[C]. International Conference on Learning Representations(ICLR), 2017, 1-2. [26] Henmi T, Zara E, Hirohashi Y, et al. Adaptive signal variances: CNN initialization through modern architectures[C].//International Conference on Learning Representations(ICLR), 2020, 1-22. [27] 雷雨婷，丁学文，孙彦，等. 基于Inception-v3卷积神经网络模型的人脸识别[J]. 天津职业技术师范大学学报，2019，29(4)：49-54.Lei Yuting, Ding Xuewen, Sun Yan, et al. Face recognition based on Inception-v3 convolutional neural network model[J]. Journal of Tianjin Polytechnic Normal University, 2019, 29(4): 49-54. (in Chinese with English Abstract) [28] Kingma D, Ba J. Adam: A method for stochastic optimization[C]. International Conference on Learning Representations(ICLR), 2015, 1-15. [29] Loshchilov I, Hutter F. Decoupled weight decay regularization[C].//International Conference on Learning Representations(ICLR), 2019, 1-19. [30] Yoonsuh Jung. Multiple predicting K-fold cross-validation for model selection[J]. Journal of Nonparametric Statistics, 2017, 30(1): 10-15. [31] 贾宇霞，樊帅昌，易晓梅. 基于显著性增强和迁移学习的鱼类识别研究[J]. 渔业现代化，2020，47(1)：38-46.Jia Yuxia, Fan shuaichang, Yi Xiaomei. Fish identification based on saliency enhancement and transfer learning[J]. Fisheries Modernization, 2020, 47(1): 38-46. (in Chinese with English Abstract)