A Highly Precise and Lightweight Detection Model for Water and Oil Leakage Detection

Weiming Wu,Kun Liu,M. Saad Shakeel,Xiaochuan Liao,Ming Zeng,Wenxiong Kang

DOI: https://doi.org/10.1109/CSIS-IAC60628.2023.10364050

2023-01-01

Abstract:The safety of Electric power plants is crucial to people's livelihoods, and leakages of water and oil in power plants is a major safety hazard that requires constant supervision. The traditional methods for water and oil leakage detection in power plants are unstable and uneconomical. Although deep learning models have achieved remarkable results in various fields of research, applying it directly to the power sector without careful design can make it difficult to meet actual requirements. To address these issues, we propose a lightweight attention-guided object detection model based on improved YOLOv5. We perform a lightweight modification to the original network by embedding an attention mechanism that enhances the model's ability to integrate information across channel dimensions. The addition of a more efficient cross stage partial bottleneck with two convolutions (C2F) module allows the model to extract high-level features from the continuous gradient information of the image in a comprehen-sive fashion, enhancing the model's capabilities for water and oil leakage detection. Meanwhile, we also established a data set specific to the actual business scenario to train and test our proposed model. Our experimental results demonstrate its superiority and effectiveness. Our proposed method achieves a detection accuracy of over 93% for water and oil leakage. In comparison to the original YOLOv5, our model achieves better detection results with a smaller number of model parameters and a faster detection speed, which better meets the real-time requirements of water and oil leakage detection in power plants.