Abstract: Embodiments described herein are directed to visual anomaly detection for content displayed via multi-display systems. For instance, computing devices may provide content for display by a respective display devices of a multi-display system. Each computing device provides images of the content, along with an identifier identifying the computing device as the source of the image, to a cloud-based storage system. A cloud-based visual anomaly detection system retrieves and analyzes the images from the storage system and determines whether any visual anomalies are present therein. The analysis is performed on a per-computing device basis. For instance, the system may apply a machine-learning based detection model to an image that is specific to the computing device that generated the image based on the identifier associated with the image. Upon detecting an anomaly, an automated action is performed to remediate the cause of the anomaly.

Abstract: Embodiments described herein are directed to visual anomaly detection for content displayed via multi-display systems. For instance, computing devices may provide content for display by a respective display devices of a multi-display system. Each computing device provides images of the content, along with an identifier identifying the computing device as the source of the image, to a cloud-based storage system. A cloud-based visual anomaly detection system retrieves and analyzes the images from the storage system and determines whether any visual anomalies are present therein. The analysis is performed on a per-computing device basis. For instance, the system may apply a machine-learning based detection model to an image that is specific to the computing device that generated the image based on the identifier associated with the image. Upon detecting an anomaly, an automated action is performed to remediate the cause of the anomaly.