Journal of Electrical Engineering and Automation

Machine vision technology plays a crucial role in automated anomaly detection systems, but existing algorithms still fall short in terms of detection accuracy and real-time performance. This paper proposes an optimized solution based on deep learning, which significantly enhances the system's detection performance by improving the convolutional neural network structure, introducing attention mechanisms, and optimizing data augmentation strategies. Experimental results show that the optimized algorithm outperforms traditional methods in both detection accuracy and speed, providing strong support for the practical application of automated anomaly detection systems.

Machine vision; Automatic detection; Algorithm optimization; Deep learning; Anomaly detection