混合卷积-长短期记忆网络模型在实验室违规行为检测中的应用
青岛科技大学, 国际学院 (泰中国际橡胶学院) 科技创新实验室, 山东青岛 266061
| 摘 要： | 本研究针对实验室人员违规行为检测，提出了一种基于CLSTM模型的方法。该模型结合了卷积神经网络（CNN）和长短时记忆网络（LSTM）。模型首先利用CNN感知图像特征，并通过卷积和池化操作进行特征提取。这些提取的特征随后传送至LSTM网络模型，利用LSTM的记忆能力在时间序列中捕捉和理解违规行为的模式。通过这种结构，CLSTM模型不仅能够有效处理和分析复杂的图像数据，还能在长期记忆和短期变化之间找到平衡，提高违规行为检测的准确性和鲁棒性。模型通过卷积网络增强了图像特征的表征能力，进一步优化了LSTM模型在计算能力和训练效率上的表现。相较于传统的LSTM模型，CLSTM在训练过程中展现出更快的收敛速度。为验证模型的有效性，本文在两个不同的数据集上进行了实验。实验结果显示，所提出的CLSTM模型在性能上显著优于传统的LSTM方法，性能提升幅度分别约为7%至11%。这一结果突显了CLSTM模型在实验室违规行为检测任务中的优越性。 |
| 关 键 词： | CLSTM; 卷积网络; LSTM网络; 不当行为检测 |
| DOI： | 10.57237/j.cst.2024.03.003 |
View:
Downloads:
收稿日期: 2024-07-15
接受日期: 2024-08-06
在线出版日期: 2024-08-08
Application of Convolutional Long Short-Term Memory Network Model in Laboratory Violation Behavior Detection
Technology Innovation Lab of International College (Sino-Thai International Rubber College), Qingdao University of Science and Technology, Qingdao 266061,
| Abstract: | This study proposes a method based on the CLSTM model for detecting misconduct among laboratory personnel. The model integrates Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks. Initially, the CNN perceives image features and performs feature extraction through convolutional and pooling operations. These extracted features are subsequently fed into the LSTM network, which leverages its memory capability to capture and understand patterns of misconduct in the temporal sequence. This structure enables the CLSTM model not only to effectively process and analyze complex image data but also to strike a balance between long-term memory and short-term variations, thereby enhancing the accuracy and robustness of misconduct detection. The model enhances the representation capability of image features through the convolutional network, further optimizing the LSTM model's performance in computational efficiency and training effectiveness. Compared to traditional LSTM models, CLSTM demonstrates faster convergence during training. To validate the model's effectiveness, experiments were conducted on two different datasets. The results indicate that the proposed CLSTM model significantly outperforms traditional LSTM methods, achieving performance improvements ranging from 7% to 11%. These findings underscore the superiority of the CLSTM model in laboratory personnel misconduct detection tasks. |
| Keywords: | CLSTM; CNN; LSTM; Improper Behavior Detection |
参考文献
| [1] | Hershey S, Chaudhuri S, Ellis D P W, et al. CNN architectures for large-scale audio classification [C] // 2017 ieee international conference on acoustics, speech and signal processing (icassp). IEEE, 2017: 131-135. |
| [2] | Graves, Alex, and Alex Graves. "Long short-term memory." Supervised sequence labelling with recurrent neural networks (2012): 37-45. |
| [3] | Huang, Shu-Chun L., and Wan-En Sun. "Exploration of social media for observing improper tourist behaviors in a national park." Sustainability 11.6 (2019): 1637. |
| [4] | Wang M, Zhao Y, Wu Q, et al. A YOLO-based Method for Improper Behavior Predictions [C] // 2023 IEEE International Conference on Contemporary Computing and Communications (InC4). IEEE, 2023, 1: 1-4. |
| [5] | Khaleghi, Ali, and Mohammad Shahram Moin. "Improved anomaly detection in surveillance videos based on a deep learning method." 2018 8th Conference of AI & Robotics and 10th RoboCup Iranopen International Symposium (IRANOPEN). IEEE, 2018. |
| [6] | Lea, Colin, et al. "Temporal convolutional networks: A unified approach to action segmentation." Computer Vision–ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part III 14. Springer International Publishing, 2016. |
| [7] | Grossberg, Stephen. "Recurrent neural networks." Scholarpedia 8.2 (2013): 1888. |
| [8] | Simonyan, Karen, and Andrew Zisserman. "Two-stream convolutional networks for action recognition in videos." Advances in neural information processing systems 27 (2014). |
| [9] | Yue-Hei Ng, J., Hausknecht, M., Vijayanarasimhan, S., Vinyals, O., Monga, R., & Toderici, G. (2015). Beyond short snippets: Deep networks for video classification. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 4694-4702). |
| [10] | Tee, Wei Zhong, et al. "A Close Look into Human Activity Recognition Models using Deep Learning." 2022 3rd International Conference on Computing, Networks and Internet of Things (CNIOT). IEEE, 2022. |
| [11] | Nagrath, P., Dharana, Dwivedi, S., Negi, R., & Singh, N. (2022). Real-Time Anomaly Detection Surveillance System. In Proceedings of Data Analytics and Management: ICDAM 2021, Volume 1 (pp. 665-678). Springer Singapore. |
| [12] | Einy, S., Sen, E., Saygin, H., Hivehchi, H., & Dorostkar Navaei, Y. (2023). Local Binary Convolutional Neural Networks' Long Short-Term Memory Model for Human Embryos' Anomaly Detection. Scientific Programming, 2023. |
| [13] | Yang, J., Qu, J., Mi, Q., & Li, Q. (2020). A CNN-LSTM model for tailings dam risk prediction. IEEE Access, 8, 206491-206502. |
| [14] | Rigatti, Steven J. "Random forest." Journal of Insurance Medicine 47.1 (2017): 31-39. (RF). |
| [15] | Kalyta, O., Barmak, O., Radiuk, P., & Krak, I. (2023). Facial Emotion Recognition for Photo and Video Surveillance Based on Machine Learning and Visual Analytics. Applied Sciences, 13(17), 9890. |
