Abstract: The disclosure relates to an artificial intelligence (AI) system for mimicking functions, such as cognition and determination as of the human brain, by utilizing a machine learning algorithm such as deep learning, and an application thereof. Provided are a neural network learning method according to an AI system and applications thereof, the method including extracting, by using a masking filter having an effective value in a specific portion of the masking filter including weight information of at least one hidden layer included in a learning network model, characteristics of input data according to weight information of a filter corresponding to the specific portion, comparing output data with target data, the output data being obtained from the learning network model based on extracted characteristics of the input data, and updating a size of the specific portion having the effective value in the masking filter, based on a result of the comparing.

Abstract: A method for detecting an anchor-free object based on deep learning may include receiving a target image for object detection, inputting the image to a backbond network and performing learning to generate a heatmap, post-processing the heatmap, and outputting the post-processed result on a screen.

Abstract: The present disclosure provides a neural network training device for recognizing a class of an object included in an image based on an artificial intelligence (AI) system and an application thereof, the neural network training method including: acquiring, by using a first learning network model trained based on source training images respectively included in at least one class, feature information of a query image included in a class different from the at least one class; obtaining a generated image from the feature information of the query image by using a second learning network model acquiring feature information of the obtained generated image by using the first learning network model; and updating weights of layers respectively included in the first and second learning network models, based on a difference between the feature information of the query image and the feature information of the generated image and on a difference between the query image and the generated image.

Abstract: An apparatus for recognizing an object using an image includes a depth map generator that generates a depth map using a feature map of the image based on a dilated convolutional neural network (DCNN) and an object recognition device that recognizes the object using the depth map generated by the depth map generator and the image.

Abstract: A method for detecting an anchor-free object based on deep learning may include receiving a target image for object detection, inputting the image to a backbond network and performing learning to generate a heatmap, post-processing the heatmap, and outputting the post-processed result on a screen.

Abstract: At least some example embodiments disclose a device and a method for generating a synthetic image and a different-angled image and eliminating noise. The method may include receiving input images, extracting feature values corresponding to the input images using an image learning model, the image learning model permitting an input and an output to be identical and generating a synthetic image based on the feature values corresponding to the input images using the image learning model.

Abstract: An apparatus for recognizing an object using an image includes a depth map generator that generates a depth map using a feature map of the image based on a dilated convolutional neural network (DCNN) and an object recognition device that recognizes the object using the depth map generated by the depth map generator and the image.

Abstract: The present disclosure provides a neural network training device for recognizing a class of an object included in an image based on an artificial intelligence (AI) system and an application thereof, the neural network training method including: acquiring, by using a first learning network model trained based on source training images respectively included in at least one class, feature information of a query image included in a class different from the at least one class; obtaining a generated image from the feature information of the query image by using a second learning network model acquiring feature information of the obtained generated image by using the first learning network model; and updating weights of layers respectively included in the first and second learning network models, based on a difference between the feature information of the query image and the feature information of the generated image and on a difference between the query image and the generated image.

Abstract: The disclosure relates to an artificial intelligence (AI) system for mimicking functions, such as cognition and determination as of the human brain, by utilizing a machine learning algorithm such as deep learning, and an application thereof. Provided are a neural network learning method according to an AI system and applications thereof, the method including extracting, by using a masking filter having an effective value in a specific portion of the masking filter including weight information of at least one hidden layer included in a learning network model, characteristics of input data according to weight information of a filter corresponding to the specific portion, comparing output data with target data, the output data being obtained from the learning network model based on extracted characteristics of the input data, and updating a size of the specific portion having the effective value in the masking filter, based on a result of the comparing.

Abstract: At least some example embodiments disclose a device and a method for generating a synthetic image and a different-angled image and eliminating noise. The method may include receiving input images, extracting feature values corresponding to the input images using an image learning model, the image learning model permitting an input and an output to be identical and generating a synthetic image based on the feature values corresponding to the input images using the image learning model.

Abstract: At least some example embodiments disclose a device and a method for generating a synthetic image and a different-angled image and eliminating noise. The method may include receiving input images, extracting feature values corresponding to the input images using an image learning model, the image learning model permitting an input and an output to be identical and generating a synthetic image based on the feature values corresponding to the input images using the image learning model.

Abstract: The video-motion segmentation method using a randomized voting is provided which includes receiving the video, extracting a plurality of feature points from the video, and grouping the plurality of feature points by applying a randomized voting method using a score histogram on each of the at least some feature points of the plurality of feature points.

Abstract: Provided is an image surveillance method and apparatus. The image surveillance method includes photographing a predetermined place and generating an image signal corresponding to the predetermined place, detecting a change in the image signal by calculating a difference between a current frame and a previous frame of the generated image signal, and generating alarm information corresponding to the detected change in the image signal in order to alarm a user.

Abstract: An apparatus and a method of guiding a user along a travel path by using global positioning system (GPS) information are described. The method includes aligning images in a selected order based on a user input; extracting location information of the aligned images; setting a travel destination and a travel sequence based on the location information extracted from the aligned images; obtaining the user's location by using the global positioning system (GPS); and guiding the user along a travel path to the travel destination based on the user's location.

Abstract: The video-motion segmentation method using a randomized voting is provided which includes receiving the video, extracting a plurality of feature points from the video, and grouping the plurality of feature points by applying a randomized voting method using a score histogram on each of the at least some feature points of the plurality of feature points.

Abstract: At least some example embodiments disclose a device and a method for generating a synthetic image and a different-angled image and eliminating noise. The method may include receiving input images, extracting feature values corresponding to the input images using an image learning model, the image learning model permitting an input and an output to be identical and generating a synthetic image based on the feature values corresponding to the input images using the image learning model.

Abstract: An apparatus and a method of guiding a user along a travel path by using global positioning system (GPS) information are described. The method includes aligning images in a selected order based on a user input; extracting location information of the aligned images; setting a travel destination and a travel sequence based on the location information extracted from the aligned images; obtaining the user's location by using the global positioning system (GPS); and guiding the user along a travel path to the travel destination based on the user's location.

Abstract: A communication system for supporting a primary user and a secondary user is provided. A base station of a communication system, includes a scheduler to group at least one primary user and at least one secondary user corresponding to each of the at least one primary user according to characteristics of services desired by users, a superposition coding unit to perform superposition coding for primary data associated with the at least one primary user and secondary data associated with the at least one secondary user to generate a transmission data stream, and a beamformer to perform beamforming for the transmission data stream.

Abstract: Provided is an image processing method and apparatus for improving image quality and enhancing a three-dimensional effect. By extracting an interest feature index from an input image based on a depth sensation estimation cue existing in the image, calculating an interest feature index integration map based on the interest feature index and importance of interest of the depth sensation estimation cue, and performing discrimination processing of the image based on the interest feature index integration map, three-dimensional effect and true sensation enhancement processing can be performed based on the probability of the existence of a depth sensation estimation cue, without segmentation of an object of interest, correct depth information, or an object model.

Abstract: Provided is an image processing method and apparatus. The image processing method includes receiving an image signal an image, detecting a feature of the image, and analyzing the image signal based on the detected feature of the image, performing segmentation on the image signal according to the analysis result, and performing image processing on the image signal according to the segmentation result.