Abstract: There is a method for monitoring emotion and behavior occurring during a conversation with a user who is in need of protection. According to an embodiment, a user monitoring method receives an input of a user utterance and generates a conversation content by using a conversation model based on a user utterance content, converts the generated conversation content into a voice and outputs the voice, and recognizes emotion of the user by using an emotion analysis model based on the user utterance content and monitors the emotion. Accordingly, the method may monitor behavior/emotion of a user through a natural conversation while allowing a specialist/guardian to monitor an abnormal sign of a user who is in need of protection.

Abstract: A system evaluates residential satisfaction of residents living in a building, based on personal tendencies of the residents and sensing data collected through sensors in the building. The system may include a server that evaluates the residential satisfaction of the residents by analyzing a correlation between the sensing data and residential satisfaction information obtained from the survey conducted through a communication manner including the user terminals. The server may control an air conditioning system in the building of a control target by predicting a feature-based preferred section of the users through random forest model learning.

Abstract: There is a method for applying a BMC analytical local fan control model in a rugged environment. A server chassis cooling fan control method according to an embodiment controls rotation speeds of cooling fans on a zone basis while identifying/managing a temperature distribution of an edge server chassis on a zone basis through BMC data analysis. Accordingly, a damage that may be caused by increased temperature of an edge server in a rugged environment may be minimized, and also, power consumption for cooling an edge server may be reduced.

Abstract: There is a method for monitoring emotion and behavior occurring during a conversation with a user who is in need of protection. According to an embodiment, a user monitoring method receives an input of a user utterance and generates a conversation content by using a conversation model based on a user utterance content, converts the generated conversation content into a voice and outputs the voice, and recognizes emotion of the user by using an emotion analysis model based on the user utterance content and monitors the emotion. Accordingly, the method may monitor behavior/emotion of a user through a natural conversation while allowing a specialist/guardian to monitor an abnormal sign of a user who is in need of protection.

Abstract: There are provided a system and a method for setting a synchronization agent. A system for setting an optimal synchronization agent according to a workload according to an embodiment includes: a data integration device configured to integrate and store data which is collected through a collection agent; an agent management server configured to manage the collection agent that periodically polls data from a data source and transmits the data to the data integration device; and an agent setting automation device configured to set an optimal agent set value related to collection and transmission of data of the collection agent.

Abstract: There is provided a cooperative driving control method based on a vehicle security status. The cooperative driving method according to an embodiment includes: identifying a security status of a cooperative driving target vehicle; determining a cooperative driving strategy according to the identified security status; and controlling driving according to the determined cooperative driving strategy. Accordingly, cooperative driving may be performed or excluded based on a security status of a target vehicle, so that an accident may be prevented from being caused by cooperative driving with a problematic vehicle, and safety of a driver and an occupant may be guaranteed.

Abstract: There are provided a method and a system for selecting an optimal synchronization broker for each workload. A system for selecting an optimal synchronization broker for each workload according to an embodiment includes: a storage unit including a database that contains specialist recommendation information including information on a correct protocol that is recommended by a specialist with respect to a service specification and data transmission requirements; and a processor configured to determine an optimal in-bound protocol and an optimal out-bound protocol according to the data transmission requirements through a decision tree model that is trained by using the specialist recommendation information as training data.

Abstract: Disclosed is a method of predicting a block vector in a multi-view image, the method including allowing an encoder to determine a reference block of a reference image for a specific block of a depth image or a specific-view image among multi-view images, allowing the encoder to determine block vector information for intra block copy (IBC) of the specific block on the basis of a block vector of the reference block when the reference block is encoded using the IBC, and allowing the encoder to encode the block vector information of the specific block. The reference image of the specific view is an image of another view, and the reference image of the depth image is a texture image corresponding to the depth image.

Abstract: Provided is a holographic optical element printing method using a tunable focus lens and a rotating mirror. According to an embodiment, a holographic printer includes: a first optical engine and a second optical engine configured to adjust a phase of an incident collimated beam and emit the collimated beam; and a first reduction optical system and a second reduction optical system configured to reduce the beam emitted from the first optical engine and the second optical engine and to allow the beam to enter a holographic material, wherein each of the first optical engine and the second optical engine includes: a rotating mirror configured to reflect while adjusting the phase of the incident collimated beam through rotation; and a tunable focus lens configured to refract while adjusting the phase of the incident collimated beam reflected from the rotating mirror through focus tuning.

Abstract: Provided is a cognitive experiment method for change in periphery region image quality for parameter determination of a foveated hologram. According to an embodiment, a cognitive experiment method for determining foveated image parameters includes: generating a foveated image which includes a foveal region and a periphery region while reducing image quality regarding the periphery region; displaying the generated foveated image; receiving an input of a response to image quality of the periphery region from a subject; and collecting information regarding appropriate image quality of the periphery region, based on the response inputted by the subject. Accordingly, in applying a foveated hologram generation algorithm for real-time reproduction, various parameters such as a boundary between a foveal region image and a periphery region image in a foveated hologram, and a degree of quality degradation of the periphery region image may be appropriately determined through a cognitive experiment.

Abstract: An image acquisition apparatus for acquiring an in-game 360 virtual reality (VR) image by using a plurality of virtual cameras includes a virtual camera group taking in-game images by using the plurality of virtual cameras. The image acquisition apparatus may also include a renderer generating textures for the taken images, generating a panoramic image by performing an equirectangular projection (ERP) mapping of the generated textures, and encoding the generated panoramic image. The image acquisition apparatus may further include an image generator generating a 360 VR image by combining audio information with the encoded panoramic image.

Abstract: The present disclosure relates to an apparatus and method for fast refining segmentation for a V-PCC encoder. The apparatus may include a grid segmentation unit segmenting a coordinate space of a point cloud into grid units, and an edge cube search unit searching a cube containing one or more points among the cubes segmented into grid units and containing a segment boundary. The apparatus may also include a surrounding cube search unit searching an edge surrounding cube containing one or more points within a predetermined range from the edge cube, and a smooth score calculation unit calculating smooth scores for all the edge surrounding cubes and all the edge cubes. The apparatus may further include a projection plane index update unit obtaining a normal score based on the calculated smooth scores and updating a projection plane index of each point in the edge cube using the normal score.

Abstract: Provided is a hologram image normalization method for a holographic printer. In a holographic printing method according to an embodiment, generating, encoding, and normalizing for the (n+1)-th hogel are performed in parallel with loading and recording of a normalized hologram for the n-th hogel, and moving and waiting for the (n+1)-th hogel. Accordingly, a global maximum value and a global minimum value for normalization may be calculated as approximate estimation values, and a hologram generation process and a printing process may be performed in parallel, so that a total printing time may be minimized and memory usage may be optimized when holographic printing is performed.

Abstract: Provided are an IoT device and an IoT communication method. The IoT device may: be connected to a server via NAT; execute a first session for service connection; and execute a second session for calculating a particular keep-alive value corresponding to a period in which an IP address mapped by the NAT expires. The present invention can minimize transmission of a keep-alive value of an IoT device, enables an IoT device to share a particular keep-alive value with other IoT devices, and thus can minimize a network load which may be caused by a keep-alive value of an IoT communication device.

Abstract: Provided is a method for regenerating a tool path on the basis of output data feedback in order to improve 3D printing output reliability. A tool path regeneration system according to an embodiment of the present invention comprises: a slicing unit for configuring a process parameter for 3D printing and performing slicing for 3D model data on the basis of the configured process parameter to generate a job file; and an output unit for performing 3D printing on the basis of the generated job file and collecting output data that is output while the 3D printing is performed, wherein the slicing unit comprises an output data analysis module for performing monitoring on the basis of the output data received from the output unit and determining whether to correct the process parameter on the basis of the result of the monitoring.

Abstract: An on-device based data analysis system and a method are proposed. According to specific examples of the system and method, when a size of measurement data collected from sensors, processes, or various equipment according to a user request is greater than or equal to an upper limit of a predetermined range of an on-device resource usage rate, the collected measurement data is transmitted to a cloud server to perform noise removal, pre-processing, and analysis on the measurement data on a cloud-basis, and when the size of the measurement data is within the predetermined range of the on-device resource usage rate or is less than a lower limit, the noise removal, pre-processing, and analysis is performed on the measurement data on an on-device basis, so that distributed processing between the on-device processing device and the cloud server is performable, whereby loads on the cloud server may be reduced.

Abstract: There is provided a video processing method performed by a computing device, the method including the steps of: collecting video from an external device; generating preprocessed data by extracting two-dimensional or three-dimensional skeleton information from the video; pre-training a first artificial intelligence model including N transformer blocks from the preprocessed data by applying an attention from a body of an object to a plurality of joints, an attention from each of the plurality of joints to the body, and an attention between persons; and learning, when parameters determined as a result of the pre-training of the first artificial intelligence model are transferred, a method of recognizing an action from the video received from the external device, using a second artificial intelligence model including the N transformer blocks on the basis of the parameters, wherein N is a natural number equal to or larger than 2.

Abstract: A pellicle for extreme ultraviolet (EUV) lithography is based on yttrium carbide and used in a EUV lithography process. The pellicle for EUV lithography includes a pellicle layer that has a core layer containing yttrium carbide. The yttrium carbide is YCx in which the atomic percentage of carbon is within a range of 25% to 45%.

Abstract: The gas sensor according to an exemplary embodiment of the present invention comprises: an FET device including one or more gate electrodes; a sensor array part including a plurality of sensors, in which a first electrode of each sensor is connected to at least one gate electrode of the plurality of gate electrodes in the FET device; and a controller detecting a gas using a current between a drain-source in response to voltage changes in the gate electrode of the FET device, wherein each sensor includes: a first electrode connected to a gate electrode of the FET device; a second electrode receiving an operating voltage through a switch controlled by the controller; and a detection film interposed between the first electrode and the second electrode.

Abstract: Provided are a hyperparameter optimizer and method for optimizing hyperparameters and a spiking neural network processing unit. The optimizer includes a statistical analyzer configured to receive training data and perform statistical analysis on the training data, an objective function generator configured to generate hyperparameter-specific objective functions by using a statistical analysis value of the statistical analyzer, and an optimal hyperparameter selector configured to select optimal hyperparameters according to certain rules on the basis of the hyperparameter-specific objective functions.