Visual anomaly detection by distributed deep learning
Ruiguang Hu, Peng Sun, Yifan Ge · AOPC 2020: Optical Sensing and Imaging Technology · 2020
Anomaly detection with visual information by distributed deep learning is proposed in the paper. First, visual anomalies are defined in a special application domain, which are very important and critical for safe operation. Secondly, deep convolutional neural network is chosen as detector for visual anomalies. Thirdly, detection results from different visual sources are fused to get higher accuracies and lower false alarm rate. Experimental results demonstrate that the visual anomaly detection framework proposed can achieve high performance and provide satisfactory security assurance.