化学分析实验中的行为识别研究
Research on Action Recognition in Chemical Analysis Experiments
DOI: 10.12677/CSA.2022.124121, PDF,    国家科技经费支持
作者: 李晓旭, 马兴录, 孙 昊：青岛科技大学，信息科学技术学院，山东 青岛
关键词: 动作识别；光流；双流卷积；融合；Action Recognition； Optical Flow； Two-Stream Convolution； Fusion
摘要: 为了提高化学实验室的智能化水平，本文对双流卷积网络进行研究，提出一种应用在化学分析实验领域的行为识别网络。本文网络将输入视频分为RGB和光流，通过改进的EfficientNetv2网络加深网络结构提取特征信息，同时探究双流网络的融合位置，改进损失函数，确定网络的最佳策略。经验证在自制的数据集中改进的双流卷积神经网络模型准确率可以达到92.4%，相比常规的行为识别算法具有更高的识别率。
Abstract: In order to improve the intelligence level of the chemical laboratory, this paper studies the two-stream convolutional network and proposes an action recognition network applied in the field of chemical analysis experiments. In this paper, the network divides the input video into RGB and optical flow, and uses the improved EfficientNetv2 network to deepen the network structure to extract feature information. At the same time, it explores the fusion position of the dual-stream network, improves the loss function, and determines the best strategy for the network. It has been verified that the accuracy rate of the improved two-stream convolutional neural network model in the self-made dataset can reach 92.4%, which is higher than the conventional behavior recognition algorithm.
文章引用：李晓旭, 马兴录, 孙昊. 化学分析实验中的行为识别研究[J]. 计算机科学与应用, 2022, 12(4): 1192-1201. https://doi.org/10.12677/CSA.2022.124121

参考文献

[1] 王春安, 危紫翼, 杨茜, 何宏斌, 徐艺珺, 陈艳萍, 曾坚朋. 国外先进实验室人员配置与经费情况对我国实验室建设运行的启示[J]. 实验技术与管理, 2021, 38(12): 243-248+282.
[2] 冯云华. 智能化学化工实验室的研究与设计[J]. 石化技术, 2019, 26(3): 310+309.
[3] 何爱欣. 桌面机械臂智能交互技术研究及应用[D]: [硕士学位论文]. 青岛: 青岛科技大学, 2019
[4] 王涛. 基于ROS的多桌面型机械臂协作机制研究与应用[D]: [硕士学位论文]. 青岛: 青岛科技大学, 2021.
[5] 张兴强. 桌面型机械臂在化学实验中的应用研究[D]: [硕士学位论文]. 青岛: 青岛科技大学, 2021.
[6] 梁绪, 李文新, 张航宁. 人体行为识别方法研究综述[J]. 计算机应用研究, 2022, 39(3): 651-660.
[7] 郝树新, 林锦州, 刘芳. 机器视觉在汽车质量检验领域的应用研究[J]. 时代汽车, 2021(21): 16-17.
[8] 袁依文, 雷斌. 无人机视觉技术在道路裂缝分类中的应用[J/OL]. 机械设计与制造: 1-6. 2021-11-08.[CrossRef] 
[9] 李柯泉, 陈燕, 刘佳晨, 牟向伟. 基于深度学习的目标检测算法综述[J/OL]. 计算机工程: 1-17. 2021-11-17.[CrossRef] 
[10] 施徐伟, 丁书艺, 储妤, 郑思婕, 韦朴. 基于机器视觉的垃圾分拣系统[J]. 电子测试, 2021(21): 35-36+42.
[11] 凌远坤, 仲志丹, 李栋, 金容鑫, 姚晨辉, 王继茂. 基于机器视觉图像识别及处理的方法[J]. 科技风, 2019(15): 95.
[12] Zhao, R., Ali, H. and Patrick, V. (2017) Two-Stream RNN/CNN for Action Recognition in 3D Videos. 2017 IEEE/RSJ International Conference on In-telligent Robots and Systems, Vancouver, 24-28 September 2017, 4260-4267. [Google Scholar] [CrossRef] 
[13] 陈颖, 来兴雪, 周志全, 秦晓宏, 池亚平. 基于3D双流卷积神经网络和GRU网络的人体行为识别[J]. 计算机应用与软件, 2020, 37(5): 164-168+218.
[14] 谢佳龙, 张波涛, 吕强. 一种基于双流融合3D卷积神经网络的动态头势识别方法[J]. 电子学报, 2021, 49(7): 1363-1369.
[15] 曾明如, 郑子胜, 罗顺. 结合LSTM的双流卷积人体行为识别[J]. 现代电子技术, 2019, 42(19): 37-40.
[16] Feichtenhofer, C., Pinz, A. and Zisserman, A. (2016) Convolutional Two-Stream Network Fusion for Video Action Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 27-30 June 2016, 1933-1941. [Google Scholar] [CrossRef] 
[17] Tan, M. and Le, Q.V. (2021) EfficientNetV2: Smaller Models and Faster Training. 2021 International Conference on Machine Learning, Online, 18-24 July 2021.
https://arxiv.org/abs/2104.00298
[18] Pakzad, F., Portmann, M., Tan, W.L. and Indulska, J. (2014) Efficient Topology Discovery in Software Defined Networks. 2014 8th International Conference on Signal Processing and Communication Systems (ICSPCS), Gold Coast, 15-17 December 2014, 1-8. [Google Scholar] [CrossRef] 
[19] Hong, L., Thanh, N.C. and Long, T.Q. (2020) Polyp Seg-mentation in Colonoscopy Images Using Ensembles of U-Nets with EfficientNet and Asymmetric Similarity Loss Func-tion. 2020 RIVF International Conference on Computing and Communication Technologies (RIVF), Chi Minh City, 14-15 October 2020, 1-6. [Google Scholar] [CrossRef] 
[20] Tan, M. and Le, Q.V. (2019) EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. 2019 International Conference on Machine Learning, Long Beach, 10-15 June 2019, 6105-6114.
[21] Yu, F. and Koltun, V. (2015) Multi-Scale Context Aggregation by Dialate Convolu-tions. arXivpreprint ar Xiv:1511.07122.
[22] Wang, F., Cheng, J., Liu, W.Y. and Liu, H. (2018) Additive Margin Softmax for Face Verification. IEEE Signal Processing Letters, 25, 926-930. [Google Scholar] [CrossRef] 
[23] Wen, Y.D., Zhang, K.P., Li, Z.F. and Qiao, Y. (2016) A Dis-criminative Feature Learning Approach for Deep Face Recognition. Proceedings of the European Conference on Com-puter Vision, Amsterdam, 8-14 September 2018, 499-515. [Google Scholar] [CrossRef] 
[24] Farnebck, G. (2003) Two-Frame Motion Estimation Based on Polynomial Expansion. 13th Scandinavian Conference on Image Analysis (SCIA 2003), Halmstad, 29 June-2 July 2003, 363-370. [Google Scholar] [CrossRef] 
[25] Farnebck, G. (2002) Polynomial Expansion for Orienta-tion and Motion Estimation. Linköping University, Linköping.