Abstract: Disclosed are a deep neural network lightweight device based on batch normalization, and a method thereof. The deep neural network lightweight device based on batch normalization includes a memory that stores at least one data and at least one processor that executes a network lightweight module. When executing the network lightweight module, the processor performs learning on an input neural network based on sparsity regularization to adaptively determine at least one parameter of the sparsity regularization, performs pruning on the learning result, and performs fine tuning on the pruning result.

Abstract: Provided are a method and system for training a dynamic deep neural network. The method for training a dynamic deep neural network includes receiving an output of a last layer of the deep neural network and outputting a first loss, receiving an output of a routing module according to an input class of the deep neural network and outputting a second loss, calculating a third loss based on the first loss and the second loss, and updating a weight of the deep neural network by using the third loss.

Abstract: An apparatus and method for searching a neural network architecture may be disclosed. The apparatus may include an architecture searcher and an architecture evaluator. The architecture searcher may search for a topology between nodes included in a basic cell of a network, search for an operation to be applied between the nodes after searching for the topology, and determine the basic cell. The architecture evaluator may evaluate performance of the determined basic cell.

Abstract: An apparatus for object detection according to an example includes a level image generating unit configured to generate a plurality of level images with reference to a target image; a feature vector extracting unit configured to extract a feature vector from each level image; a codeword generating unit configured to generate a codeword by clustering the feature vector for each level image; a histogram generating unit configured to generate a histogram corresponding to the codeword; and a classifier configured to generate object recognition information of the target image based on the histogram.

Abstract: Disclosed herein is an apparatus and method of searching for a wanted vehicle, capable of interoperating with black boxes mounted in vehicles of unspecified individuals, recognizing and searching for registration numbers of vehicles in proximity of each black box in real time, and identifying a location of the wanted vehicle in real time using information about locations of the searched vehicles. The method includes requesting, by an apparatus for searching for a wanted vehicle, a black box installed in at least one vehicle to search for a registration number of the wanted vehicle, and receiving a response corresponding to the request, and acquiring information about the wanted vehicle and a location of the wanted vehicle corresponding to the response using the black box for recognizing a vehicle registration number or the black box for detecting a vehicle registration number region.

Abstract: A human tracking method using a color histogram is disclosed. The human tracking method using the color histogram according to the present invention can more adaptively perform human tracking using different target color histograms according to the human poses, instead of applying only one target color histogram to the tracking process of one person, such that the accuracy of human tracking can be increased. The human tracking method includes performing color space conversion of input video data; calculating a state equation of a particle based on the color-space conversion data; calculating the state equation, and calculating human pose-adaptive observation likelihood; resampling the particle using the observation likelihood, and estimating a state value of the human; and updating a target color histogram.

Abstract: Disclosed herein is an apparatus and method of searching for a wanted vehicle, capable of interoperating with black boxes mounted in vehicles of unspecified individuals, recognizing and searching for registration numbers of vehicles in proximity of each black box in real time, and identifying a location of the wanted vehicle in real time using information about locations of the searched vehicles. The method includes requesting, by an apparatus for searching for a wanted vehicle, a black box installed in at least one vehicle to search for a registration number of the wanted vehicle, and receiving a response corresponding to the request, and acquiring information about the wanted vehicle and a location of the wanted vehicle corresponding to the response using the black box for recognizing a vehicle registration number or the black box for detecting a vehicle registration number region.

Abstract: A human tracking method using a color histogram is disclosed. The human tracking method using the color histogram according to the present invention can more adaptively perform human tracking using different target color histograms according to the human poses, instead of applying only one target color histogram to the tracking process of one person, such that the accuracy of human tracking can be increased. The human tracking method includes performing color space conversion of input video data; calculating a state equation of a particle based on the color-space conversion data; calculating the state equation, and calculating human pose-adaptive observation likelihood; resampling the particle using the observation likelihood, and estimating a state value of the human; and updating a target color histogram.

Abstract: The present invention provides an image separation apparatus and method. The image separation apparatus includes an image reception unit for receiving an input image. A background model generation unit generates a background model corresponding to the input image. A foreground/background separation unit performs a task of determining using the background model whether reference pixels among all pixels of the input image belong to a foreground or a background, and performs a task of estimating, based on results of the foreground/background determination task, whether remaining pixels other than the reference pixels belong to the foreground or the background.

Abstract: A method of authoring experiential learning content includes displaying an authoring window to author the experiential learning content when a request is made to author the experiential learning content; and creating a virtual world by loading and arranging 3D objects and a 2D objects, in the authoring window, which corresponds to a scenario of the experiential learning content. Further, the method includes defining an Action-zone that determines a position where a user is merged in the virtual world; and defining a state by dividing the scenario into a plurality of steps as time goes by based on the scenario to play the experiential learning content from a specific time point, respectively. Furthermore, the method includes defining a processing routine of an event occurred in the state; and authoring the experiential learning content according to the defined Action-zone, the defined state, and the defined processing routine.

Abstract: An educational system using virtual reality includes an image computer for extracting an image of a student from an entire image acquired by capturing the student; a remote lecturer computer for extracting an image of a lecturer at a remote place from an entire image acquired by capturing the remote lecturer. Further, the educational system using the virtual reality includes a classroom lecturer computer for executing learning content in which a virtual space is implemented, and performing real-time communication with the remote lecturer computer; and a virtual reality computer for combining the image of the student and the image of the remote lecturer with the virtual space of the learning content.

Abstract: A multi-image acquisition apparatus for use in a biometrics system includes a light splitting filter for reflecting or transmitting multi-image lights representing an image of a target subject provided along an identical light path into different bands; a first image sensor for imaging an iris area by capturing a fraction of the multi-image light which has been transmitted through the light splitting filter; and a second image sensor for imaging a facial area by capturing a fraction of the multi-image light which has been reflected from the light splitting filter. The apparatus further includes a control module for controlling the first image sensor and the second image sensor and providing the images obtained by the first sensor and second sensor for image recognition.

Abstract: Disclosed herein is a method of providing a trusted path between a client and a system using an access control processing technique. The method of providing a trusted path between a client and a system includes the step of determining whether access to resources of the system will be permitted or refused on the basis of access control rules and databased attributes set by a security administrator. Thereafter, the client is notified of permission for or refusal of the access in accordance with the result of the determination.

Abstract: A file security system uses a security class set by an access control module. The file security system includes a disk, a kernel memory and an encryption file system. The disk includes a key file in which an encryption key corresponding to the security class is stored and a file encoded by the encryption key. The encryption key stored in the disk is loaded into the kernel memory when the file security system starts operating. The encryption file system extracts an encryption key corresponding to a security class of a file that a user intends to read or store; decodes or encodes the file by using the extracted encryption key; and then provides the decoded file to the user or stores the encoded file in the disk.

Abstract: Disclosed herein is a method of providing a trusted path between a client and a system using an access control processing technique. The method of providing a trusted path between a client and a system includes the step of determining whether access to resources of the system will be permitted or refused on the basis of access control rules and databased attributes set by a security administrator. Thereafter, the client is notified of permission for or refusal of the access in accordance with the result of the determination.