Electronics and Telecommunications Research Institute

Jin Young Moon

Hyung Il Kim

Ki Min Yun

Yong Jin Kwon

Jong Youl Park

Jeun Woo Lee

Sung Chan Oh

Kang Min Bae

Yongjin Kwon

Byoung Youl Song

Chang Seok Bae

RockWon Kim

Sungchan Oh

Jongyoul Park

Daeha Lee

Seung Woo Jung

Kimin Yun

Joon Young Jung

Jong Hee Kim

Hyun Ki Kim

Young Jo Cho

Jungsook Kim

Hyun Kyu Cho

Hyung Jik Lee

Chan Kyu Park

Kyu Chang Kang

Dong Oh Kang

Jang Ho Choi

Yu Seok Bae

Kyung Il Kim

Hyeonsung Cho

Yeon Jun Kim

Moonyoung Chung

Ho Sang Ham

Youngwan Lee

Hyungil Kim

Young Rae Kim

Yoon Joon Lee

Seunglak Choi

Je Seok Ham

Joong Won Hwang

Hyukjae Jang

Jung Kyoo Shin

Hangkyu Kim

Dong Jin Sim

Disclosed herein are a method and apparatus for processing feature information based on an artificial neural network. According to an embodiment of the present disclosure, the apparatus for processing feature information based on an artificial neural network may include a memory for storing data and a processor for controlling the memory, and the processor may further be configured to extract a graph, which includes vertices, based on a feature map of an image, to extract a feature vector corresponding to the vertices and to process the graph and the feature vector based on an artificial neural network, and the graph may include positions of the vertices and information on a connection relationship between the vertices.

Method and apparatus for feature transforming and processing based on artificial neural network

Provided are an apparatus and method for recognizing whether an action objective is achieved. The apparatus includes a video feature extraction module configured to receive a video and output a video feature sequence through an operator, such as a convolution operator, an action case memory module configured to compress and store action case information for each action type and each of success and failure groups transmitted from the video feature extraction module and return the action case information according to a query including a pair of an action type and whether the action is successful, and an action success or failure determination module configured to receive a pair of a video feature sequence and an action type identifier and output whether an action of the given video feature sequence is successful in connection with the action case memory module.

Apparatus and method for recognizing whether action objective is achieved

A method and apparatus for face recognition robust to an alignment of the face comprising: estimating prior information of a facial shape from an input image cropped from an image including a face using the first deep neural network (DNN); extracting feature information of facial appearance from the input image by using a second DNN; training, by using a face image decoder based on the prior information and the feature information, the face recognition apparatus; and extracting, from a test image, facial shape-aware features in the inference step by using the trained second DNN.

Method and apparatus for face recognition robust to alignment status of the face

Disclosed herein are a method and apparatus for searching for a video section by using a natural language. The method for searching for a video section includes: extracting keywords from a natural language sentence, when the natural language sentence is input; determining whether or not the extracted keywords are included in predefined context information; and deriving and providing a final search result. In addition, when the extracted keywords are included in the predefined context information, a search result is derived by performing a first method, and when the extracted keywords are not included in the predefined context information, a search result is derived by performing a second method.

Method and apparatus for searching video section using natural language

Provided is a method of detecting a semantics section in a video. The method includes extracting all video features by inputting an inputted video to a pre-trained first deep neural network algorithm, extracting a query sentence feature by inputting an inputted query sentence to a pre-trained second deep neural network algorithm, generating video-query relation integration feature information in which all of the video features and the query sentence feature have been integrated by inputting all of the video features and the query sentence feature to a plurality of scaled-dot product attention layers, and estimating a video segment corresponding to the query sentence in the video based on the video-query relation integration feature information.

Method and system for retrieving video segment by a semantic query

Provided is an apparatus for online action detection, the apparatus including a feature extraction unit configured to extract a chunk-level feature of a video chunk sequence of a streaming video, a filtering unit configured to perform filtering on the chunk-level feature, and an action classification unit configured to classify an action class using the filtered chunk-level feature.

Apparatus and method for online action detection

An electronic device according to an embodiment disclosed herein may include a memory including at least one instruction and a processor. By executing the at least one instruction, the processor may check feature information corresponding to a video and including at least one of an appearance-related feature value and a motion-related feature value from the video, calculate at least one of a starting score related to a starting point of an action instance, an ending score related to an ending point of an action instance, and a relatedness score between action instances on the basis of the feature information corresponding to the video, the action instances being included in the video, and generate an action proposal included in the video on the basis of the at least one score.

Electronic device, action instance generation method, and recording medium

A domain adaptation-based object recognition apparatus includes a memory configured to store a domain adaptation-based object recognition program and a processor configured to execute the program. The processor learns a generative model for generating a feature or an image similar to a gallery image on the basis of domain adaptation in association with an input probe image and learns an object recognition classification model by using a learning database corresponding to the gallery image and the input probe image, thereby performing object recognition using the input probe image.

Domain adaptation-based object recognition apparatus and method

A method and apparatus for detecting a garbage dumping action in real time on a video surveillance system are provided. A change region, which is a motion region, from an input image is detected, joint information including joint coordinates corresponding to a region in which joints exist is generated, and an object held by a person from the image using the change region and the joint information is detected. Then, an action of dumping the object based on a distance between the object and the joint coordinates is detected.

Method and apparatus for detecting a garbage dumping action in real time on video surveillance system

Provided is a dynamic object detecting technique, and more specifically, a system and method for determining a state of a motion of a camera on the basis of a local motion estimated on the basis of a video captured by a dynamic camera and a result of analyzing a global motion, flexibly updating a background model according to the state of the motion of the camera, and flexibly detecting a dynamic object according to the state of the motion of the camera.

System and method for detecting dynamic object

Provided is a question answering system with respect to a natural language question and a method thereof. The question answering system includes a candidate answer generating unit configured to extract a document mapped to an input natural language question, and generate candidate answers with respect to the natural language question from the extracted document, a text entailment recognizing unit configured to generate a text entailment recognition result representing a degree of association between multiple evidence sentences including the generated candidate answers and the natural language question, a list generating unit configured to generate a candidate answer list including the multiple evidence sentences in high association degree order on the basis of the text entailment recognition result, and an output unit configured to output the generated candidate answer list as a search result with respect to the natural language question.

Question answering system and method

Disclosed are a smart learning system using a device social relation and a method thereof. The smart learning system includes a first device that generates a device social relation and shares learning content, and at least one second device that executes an event related to the learning content by acquiring the shared learning content when is included in the device social relation.

Smart learning system using device social relation and method thereof

Provided is a path-token-based web service caching method including determining whether or not stored cache data exists when a web service call request exists, and when the cache data does not exist, creating a predetermined path-token set and a predetermined tag data set based on a message schema of Web Services Description Language (WSDL), and creating a request Simple Object Access Protocol (SOAP) message, creating a request SOAP message template by using a path-token for the created request SOAP message, and calling the web service, and creating cache data including the tag data set, input values set, the request SOAP message template, the request SOAP message, and SOAP binding information. Accordingly, the method can solve the problems of a conventional web service caching method whereby the method can not cope with change in the number of inputs, and an exact input position is not searched for when an input value is changed. Therefore, it is possible to implement an efficient web service call environment by web service caching.

Path-token-based web service caching method

A system and method for aborting web services automatically. An application program container has a first web service application program embedded, intercepts a message that is transmitted and received between a second web service program located externally and the first web service application program, identifies the kind of an operation requested by the second web service application program, and keeps information required for compensation. A protocol manager generates and manages a first sub-transaction object by a request of the second web service application program. A local compensator compensates the first sub-transaction if a global transaction coordinator sends a command to compensate a second sub-transaction already committed. A protocol service unit transmits and receives messages for managing the first sub-transaction. A registration service unit registers the generated first sub-transaction with the global transaction coordinator.

System for aborting web services automatically and method thereof

Growing community of inventors

Daejeon, South Korea

Jin Young Moon