2results about How to "Automatic processing" patented technology

The application discloses an operation and maintenance video abnormal behavior automatic auditing method based on artificial intelligence, and specifically comprises the following steps: collecting operation and maintenance field videos in real time through proxy software, and carrying out pretreatment such as denoising, background removal and image enhancement to improve the accuracy of target detection and behavior recognition; the behavior target of operation and maintenance personnel in the video is identified by using a YOLO target detection model to obtain boundary box coordinates, category labels and confidence scores; the detection result is transmitted to a Kalman filter for continuous tracking of the target to generate motion trajectory data; 3D convolutional neural network (3D-CNN) is used to analyze space-time features, judge and label abnormal behaviors; finally, the labeled information is integrated with the video data, video metadata is embedded through FFmpeg, the quick jump function of the labeled information is realized, the audit personnel are helped to quickly locate and examine abnormal behaviors, and the auditing efficiency and the intelligent level of operation and maintenance management are greatly improved.

This invention discloses an enterprise procurement supply chain collaborative management system, comprising: a perception layer deployed at each node of the procurement supply chain, used to collect data on the flow and processing of renewable resource materials through passive RFID tags built into heterogeneous devices, and to perform data verification and synchronization through an inter-device self-organizing network; a trust layer used to bind the operator's biometric information with their identity, operating permissions, and initial credit rating, and to dynamically manage credit and cross-enterprise permissions based on their operating behavior and device feedback; and an analysis and decision-making layer used to analyze and issue early warnings on the perceived data through an edge computing module, and to simulate the value transformation path under different working conditions through a virtual mapping model. By deploying computing modules at the network edge, this invention can process data locally, independently and quickly identify and issue early warnings for anomalies such as equipment failures and quality deviations, achieving a leap from post-event response to pre-event prediction and in-event intervention.