Abstract: Incoming sensor data from a data collection device may be received at the data processing platform that includes the plurality of data processing microservices. A data processing microservice of the data processing platform may detect that the incoming sensor data from the data collection device caused an error. As a result, the incoming sensor data may be queued in a faulty data cache of the data processing platform. Subsequently, at least one of the data processing microservice or the incoming sensor data stored in the faulty data cache may be modified such that the incoming sensor data is processed by the data processing microservice without the error. Following the processing, the incoming sensor data may be deleted from the faulty data queue of the data processing platform.

Abstract: Operational states of multiple user devices may be managed based on information that is exchanged by the multiple user devices via heartbeat signals. A user device may receive heartbeat signals from multiple connected devices. A determination is made based at least on the heartbeat signals whether a higher-level hub device with an active network connection to a backend service is connected to the particular user device. Thus, in response to determining that no higher-level hub device with an active network connection to the backend service is connected to the user device, a network connection may be established via a communication channel between the particular user device and the backend service.

Abstract: In multimedia object tracking and merging of tracked objects, an object is tracked through frames of multimedia content until a frame appears in which the tracked object is not detected. A first track is designated as one or more consecutive frames in which the tracked object is detected, the first track ending at the first frame. Tracking continues to try to detect the tracked object in a second frame subsequent to the first frame. If the tracked object is not again detected, information about the first track is output. If the tracked object is detected subsequently, a second track of consecutive tracked object detection is designated. The tracked objects in the two tracks are then compared with the aid of trained data models, and a matching score is determined to reflect the degree of match. If the matching score meets or exceeds a first threshold, the compared tracks are merged using the same identifier assigned to both tracks.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, to manage device communications during emergent conditions are disclosed. In one aspect, a method includes the actions of receiving, by a first computing device, data indicating that a second computing device is configured to communicate using a first protocol. The actions include receiving data indicating that a third computing device is configured to communicate using the first protocol and a second protocol. The actions include determining that the second computing device is unable to communicate with a fourth computing device. The actions include determining a context. The actions include generating an instruction for the third computing device to communicate with the second computing device using the first protocol. The actions include outputting, to the third computing device, the instruction for the third computing device to communicate with the second computing device using the first protocol.

Abstract: This disclosure describes techniques for continuous improvement of machine learning models (also called data models) in a Content Management System (CMS). In one example, a CMS may store a set of data models for each application such as plate number recognition, facial recognition, a determination of likelihood of assault to a law enforcement officer in a traffic violation or robbery scenario, and car identification. In an example embodiment, a predictive model may be used to select a data model from the plurality of data models. The selected data model may be further improved or trained to a new sample of data features to generate an output pattern (e.g., likelihood of assault to a law enforcement officer).

Abstract: This disclosure describes techniques for continuous improvement of machine learning models (also called data models) in a Content Management System (CMS). In one example, a CMS may store a set of data models for each application such as plate number recognition, facial recognition, a determination of likelihood of assault to a law enforcement officer in a traffic violation or robbery scenario, and car identification. In an example embodiment, a predictive model may be used to select a data model from the plurality of data models. The selected data model may be further improved or trained to a new sample of data features to generate an output pattern (e.g., likelihood of assault to a law enforcement officer).

Abstract: A video file may be presented via a user application that displays one or more video frames of the video file. A user request to perform an object detection for objects of a specific object type in a video frame of the video file may be received from the user application. A machine-learning model of a plurality of machine-learning models that is configured to detect objects of the specific object type may be applied to the video frame to detect an object of the specific object type in the video frame. Each of the plurality of machine-learning models may be trained to detect objects of a corresponding object type. Subsequently, an object tracking algorithm may be applied to one or more additional video frames of the video file to track the object of the specific object type across the one or more additional video frames.

Abstract: Image readability is improved via one or more patterns taken using different light wavelengths. A first pattern may include alphanumeric characters, barcodes, Quick Response (QR) codes, or a similar unique code that can be used to identify vehicle license plates, road signals, charts, placards, advertisements, or the like, using a light wavelength such as a visible light wavelength. A paired second pattern may include a copy of the first pattern or a different pattern but constructed with a different material that responds to a different light wavelength e.g., Ultra-Violet (UV) light wavelength. In one example, the paired second pattern may be identified and used as a reference for identifying the first pattern. This technique of using multi-patterns for identifying road signals, charts, placards, advertisements, and particularly the vehicle license plates during extreme weather conditions may improve law enforcement operations or other similar purposes.

Abstract: A content management system (CMS) manages content for trained machine learning (ML) models. The CMS may develop first, second, and third trained ML models from corresponding datasets, to output respective values of dependent variables derived from data underlying the datasets as independent variables, the respective outputs having statistical confidences in the accuracy of their predictions. The third dataset results from combining the first and second datasets, and the third trained ML model is derived from training on the third dataset. The datasets and ML models are stored in a data store, with the trained ML models associated with respective datasets, the datasets with respective underlying data, the trained ML models with respective statistical confidences and corresponding thresholds, and the trained ML models with metadata indicating independent and dependent variables. The datasets and ML models can be versioned and the provenance of the datasets tracked for future ML modeling.

Abstract: A microservices platform that includes a plurality of microservices for providing data processing functions for data of multiple tenants may be implemented. Each tenant of the multiple tenants may be provided with an assigned binary large object (blob) data store in a blob storage platform that is only accessible by a corresponding tenant using a corresponding tenant identifier. Furthermore, each tenant of the multiple tenants may be assigned a corresponding local database at each microservice of the plurality of microservices that is only accessible by the corresponding tenant using the corresponding tenant identifier.

Abstract: This disclosure describes techniques for implementing an asset bucket on user devices for organizing assets in a database. Assets may include, without limitation, stored multimedia data from various sources, grouped multimedia data, events, conditions, parameters, environmental data, and other data or telemetry data that are stored in a network operating center (NOC) server database or a third-party database. The asset bucket may include a persistent working space that can be rendered as a pane on a device's user interface for organizing assets that can be selected from a rendered window or windows on the device's user interface and/or directly inputted on the persistent working space. By configuring the asset bucket to facilitate performance of actions on the selected assets, the asset bucket may improve generation of reports on these selected assets.

Abstract: A video file may be presented via a user application that displays one or more video frames of the video file. A user request to perform an object detection for objects of a specific object type in a video frame of the video file may be received from the user application. A machine-learning model of a plurality of machine-learning models that is configured to detect objects of the specific object type may be applied to the video frame to detect an object of the specific object type in the video frame. Each of the plurality of machine-learning models may be trained to detect objects of a corresponding object type. Subsequently, an object tracking algorithm may be applied to one or more additional video frames of the video file to track the object of the specific object type across the one or more additional video frames.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for automated camera activation are disclosed. In one aspect, a method includes the actions of receiving, from a sensor, sensor data. The actions further include, based on the sensor data, determining that a firearm has been removed from a holster. The actions further include, based on determining that the firearm has been removed from the holster, generating a software-recognized event that indicates the firearm has been removed from the holster. The actions further include, in response to the software-recognized event that indicates the firearm has been removed from the holster, generating an instruction for a camera to record. The actions further include providing, for output, the instruction for the camera to record.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, to dynamically utilize multiple devices are disclosed. In one aspect, a method includes the actions of receiving, by a control device, data indicating a first function of a first computing device. The actions further include receiving, by the control device, data indicating a second function of a second computing device. The actions further include determining whether the first function and the second function perform a same function. The actions further include based on a determination that the first function and the second function perform the same function and at least one goal from a user, generating an instruction based on a function selection rule to disable the second function.

Abstract: This disclosure describes techniques for implementing an asset bucket on user devices for organizing assets in a database. Assets may include, without limitation, stored multimedia data from various sources, grouped multimedia data, events, conditions, parameters, environmental data, and other data or telemetry data that are stored in a network operating center (NOC) server database or a third-party database. The asset bucket may include a persistent working space that can be rendered as a pane on a device's user interface for organizing assets that can be selected from a rendered window or windows on the device's user interface and/or directly inputted on the persistent working space. By configuring the asset bucket to facilitate performance of actions on the selected assets, the asset bucket may improve generation of reports on these selected assets.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, to manage devices during emergent conditions are disclosed. In one aspect, a method includes the actions of receiving, by a first computing device, data indicating a function of a second computing device. The actions further include determining, by the first computing device, a context of the second computing device. Based on the context of the second computing device, the actions further include determining, by the first computing device, whether to activate the function of the second computing device. Based on determining whether to activate the function of the second computing device, the actions further include determining, by the first computing device, whether to output, to the second computing device, an instruction to activate the function.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for automated camera activation are disclosed. In one aspect, a method includes the actions of receiving, from a sensor, sensor data. The actions further include, based on the sensor data, determining that a firearm has been removed from a holster. The actions further include, based on determining that the firearm has been removed from the holster, generating a software-recognized event that indicates the firearm has been removed from the holster. The actions further include, in response to the software-recognized event that indicates the firearm has been removed from the holster, generating an instruction for a camera to record. The actions further include providing, for output, the instruction for the camera to record.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, to dynamically utilize multiple devices are disclosed. In one aspect, a method includes the actions of receiving, by a first computing device, data indicating a first function of a second computing device. The actions further include receiving, by the first computing device, data indicating a second function of a third computing device. The actions further include determining, by the first computing device, that the first function of the second computing device and the second function of the third computing device perform a same function. The actions further include generating, by the first computing device, an instruction for the third computing device to disable the second function. The actions further include providing, for output by the first computing device and to the third computing device, the instruction for the third computing device to disable the second function.

Abstract: A communication broker service of a microservices platform may direct a distributed queue management service to monitor multiple data distribution outboxes of a plurality of microservices of the microservices platform for message data. The communication broker service may receive an indication from the distributed queue management service that the distributed queue management service has retrieved message data from a data distribution outbox of a first microservice. The communication broker service may apply a business logic to identify one or more second microservices that are configured to receive the message data from the first microservice. The communication broker server may then direct the distributed queue management service to distribute the message data of the first microservice from a data store of the distributed queue management service to one or more corresponding data distribution inboxes of the one or more second microservices.

Abstract: This disclosure describes techniques that enable a categorization controller to detect activation of a portable recording device that is configured to capture a real-time multimedia stream of the current event. The categorization controller may further identify a set of categories that are likely associated with the real-time multimedia stream, determine an ordered ranking of the set of categories, and generate a ranked category dataset for delivery to the portable recording device. In doing so, the portable recording device may present the ordered ranking of the set of categories at a user interface.