Abstract: Systems and methods for remotely turning on and turning up a Raman amplifier are provided. In one embodiment, a method includes the step of turning on one or more Raman pumps of a Raman amplifier to a predetermined safe gain or power level. The method also includes determining an estimated loss along a fiber optic span of a link between adjacent nodes of an optical network. Responsive to the estimated loss being greater than a reach of an Optical Supervisory Channel (OSC) signal along the link, the method includes the step of adjusting the gain or power level of the one or more Raman pumps.

Abstract: A method and device are for acquiring demand information of a component. The method including: acquiring model data information corresponding to the component; converting the model data information into component format information; and determining, based upon at least one component format information, demand information related to the component. At least one embodiment of the method and device include one of the following advantages: component information can be conveniently and automatically acquired and the component format information is automatically analyzed and classified, so that a current user demand and overall market trend are automatically determined, thus improving the accuracy of information analysis and greatly improving the efficiency for acquiring the demand information; and/or have a better universality and are applicable to various different components.

Abstract: Techniques for keypoint-based multi-label word segmentation and localization are described. A machine learning model identifies bounding regions of text within an image, and then generates multiple channel matrices representing predicted keypoints of the text within the bounding regions. The keypoints can be used to rectify the corresponding graphical content from the image including the text to improve the ability to perform optical character recognition and identify the text. Line and word segmentation and localization can be performed together.

Abstract: A surveillance system may comprise one or more computing devices and one or more robotic surveillance devices. The one or more computing devices may be configured to obtain video data captured by one or more cameras. The one or more computing devices may analyze the video data to determine whether there is any trigger event. In response to determining that there is a trigger event, the one or more computing device may determine an optimal robotic surveillance device among the one or more robotic surveillance devices based on the trigger event and provide an instruction to the optimal robotic surveillance device. The optimal robotic surveillance device may be configured to perform a responding action in response to receiving the instruction.

Abstract: An activity recognition system may comprise a local device and a server. The local device may be configured to: obtain a video; determine whether at least one human is present in one or more frames of the video; in response to determining the at least one human present in the video, determine if the at least one human in each of the frames corresponds to the same person; in response to determining that the at least one human in each of the frames corresponds to the same person, determine if the person is a stranger; and in response to determining that the person is not a stranger, transmit at least a portion of the video to the server. The server may be configured to: predict an activity class associated with the person; and trigger an alarm based on the activity class.

Abstract: An activity recognition system may comprise a local device and a server. The local device may be configured to: obtain a video; determine whether at least one human is present in one or more frames of the video; in response to determining the at least one human present in the video, determine if the at least one human in each of the frames corresponds to the same person; in response to determining that the at least one human in each of the frames corresponds to the same person, determine if the person is a stranger; and in response to determining that the person is not a stranger, transmit at least a portion of the video to the server. The server may be configured to: predict an activity class associated with the person; and trigger an alarm based on the activity class.

Abstract: A surveillance system may comprise one or more computing devices and one or more robotic surveillance devices. The one or more computing devices may be configured to obtain video data captured by one or more cameras. The one or more computing devices may analyze the video data to determine whether there is any trigger event. In response to determining that there is a trigger event, the one or more computing device may determine an optimal robotic surveillance device among the one or more robotic surveillance devices based on the trigger event and provide an instruction to the optimal robotic surveillance device. The optimal robotic surveillance device may be configured to perform a responding action in response to receiving the instruction.

Abstract: A computer-implemented method to train a video annotation machine learning process is disclosed. The method may include obtaining a video and determining that a predetermined event occurs in the video. The method may include identifying a first set of object attribute data associated with the event in the video based on a machine learning process and machine learning data. The method may include receiving a second set of object attribute data of the event in the video from a user or external source. The method may also include comparing the first set of object attribute data with the second set of object attribute data. The method may include validating the quality of the first set of object attribute data when the first set of object attribute data is determined to be sufficiently similar to the second set of object attribute data.

Abstract: A computer-implemented method to identify an object in a video is disclosed. The method may include obtaining a video and receiving an indication of one or more foreground regions in the video from a user module based on a user selection within the user module. The method may also include receiving an indication of one or more regions of interest in the foreground regions from the user module based on a user selection within the user module. The method may include determining a human potential size in a plurality of locations in the one or more regions of interest. The method may include generating a background model based on a plurality of frames of the video. The method may also include tracking an object in the one or more regions of interest based on the background model. The method may include identifying the object as a predicted human based on the human potential size.

Abstract: Systems and methods for characterizing an optical fiber performed in part by an optical node in an optical line system include performing one or more measurements to characterize the optical fiber with one or more components at the optical node, wherein the one or more components perform functions during operation of the optical node and are reconfigured to perform the one or measurements independent of the functions; and configuring the optical node for communication over the optical fiber based on the one or more measurements. The one or more components can include any of an Optical Service Channel (OSC), an Optical Time Domain Reflectometer (OTDR), and an optical amplifier. The configuring can include setting a launch power into the optical fiber based on the one or more measurements.

Abstract: A computer-implemented method to compare events from videos including monitoring two or more videos and identifying a first event in a first video, determining a first feature description of the first event, a first location of the first event, and a first time stamp of the first event, identifying a second event in a second video of the two or more videos, determining a second feature description of the second event, a second location of the second event, and a second time stamp of the second event, comparing the first and second feature description, the first and second location, and the first and second time stamp. The method may include classifying the first event and the second event as correlated events determined to be sufficiently similar and classifying the first event and the second event as not correlated events when determined not to be sufficiently similar.

Abstract: A surveillance method may comprise obtaining one or more alert levels, and for each of the alert levels, obtaining a user configuration of an action for responding to the alert level, training a model with marked training materials for detecting the alert level of a triggering event from a camera input, obtaining a user configuration of a time for executing the action, and obtaining a user configuration of a system for executing the action.

Abstract: Machine learning-based approaches are used to identify complementary sets of items, such as articles of clothing and accessories that “match,” and suggest items that would complement a given item of interest. A simulated representation of how the item of interest and the identified complementary items would look together is then generated. For example, given a particular piece of clothing or other apparel item of interest, additional items that complement the item of interest can be identified and suggested to a potential purchaser of the item. Additionally, a three-dimensional (3D) or pseudo-3D representation of a human body can be generated to model the apparel item of interest and the identified complementary apparel items to give the user an idea of how the suggested outfit would look on a user. The representation can be modified to more closely resemble a particular user.

Abstract: Disclosed in embodiments are a method and apparatus for adjusting a product model, and a storage medium. The method for adjusting a product model includes determining a life cycle data type of a product corresponding to the product model; collecting, in a life cycle of the product, a parameter value which is associated with the product and complies with the life cycle data type; and adjusting the product model based upon the parameter value collected. In an embodiment, the product model is adjusted based upon a parameter of a product life cycle type, so that the product model can reflect the real state of the product throughout the life cycle, thereby improving the accuracy of the product model.

Abstract: A surveillance system may comprise a control device and at least one robotic device. The control device is associated with a user and configured to request to connect to the at least one robotic device and in response to being connected, communicate a characteristic of the user to the at least one robotic device. The at least one robotic device comprises a platform to carry the user, and may be configured to in response to the request of the control device, verify identity of the control device of the user; in response to the identity of the control device of the user being verified, connect to the control device; define parameters of the at least one robotic device based on the characteristic of the user; and adjust the parameters of the at least one robotic device according to a riding pattern of the user.

Abstract: A surveillance system may comprise one or more computing devices and a plurality of robotic surveillance devices. The one or more computing devices may be configured to obtain video data captured by one or more cameras. The one or more computing devices may analyze the video data to identify two or more conditions, and associate results of the identification with respect to the two or more conditions to determine an activity category. The one or more computing device may assign tasks to the plurality of robotic surveillance devices based on the determined activity category. The plurality of robotic surveillance devices may be configured to perform responding actions in response to receiving the tasks.

Abstract: A computer-implemented method to train a video annotation machine learning process is disclosed. The method may include obtaining a video and determining that a predetermined event occurs in the video. The method may include identifying a first set of object attribute data associated with the event in the video based on a machine learning process and machine learning data. The method may include receiving a second set of object attribute data of the event in the video from a user or external source. The method may also include comparing the first set of object attribute data with the second set of object attribute data. The method may include validating the quality of the first set of object attribute data when the first set of object attribute data is determined to be sufficiently similar to the second set of object attribute data.

Abstract: A surveillance system may comprise one or more computing devices and one or more robotic surveillance devices. The one or more computing devices may be configured to obtain video data captured by one or more cameras. The one or more computing devices may analyze the video data to determine whether there is any trigger event. In response to determining that there is a trigger event, the one or more computing device may determine an optimal robotic surveillance device among the one or more robotic surveillance devices based on the trigger event and provide an instruction to the optimal robotic surveillance device. The optimal robotic surveillance device may be configured to perform a responding action in response to receiving the instruction.

Abstract: A surveillance method may comprise obtaining one or more alert levels, and for each of the alert levels, obtaining a user configuration of an action for responding to the alert level, training a model with marked training materials for detecting the alert level of a triggering event from a camera input, obtaining a user configuration of a time for executing the action, and obtaining a user configuration of a system for executing the action.

Abstract: An activity recognition system may comprise a local device and a server. The local device may be configured to: obtain a video; determine whether at least one human is present in one or more frames of the video; in response to determining the at least one human present in the video, determine if the at least one human in each of the frames corresponds to the same person; in response to determining that the at least one human in each of the frames corresponds to the same person, determine if the person is a stranger; and in response to determining that the person is not a stranger, transmit at least a portion of the video to the server. The server may be configured to: predict an activity class associated with the person; and trigger an alarm based on the activity class.