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 for a building on the basis of an image and a method of constructing a building search database (DB) for image-based building search. The method includes constructing a building search DB, receiving a query image from a user terminal, detecting a region to which a building belongs in the query image, extracting features of the region detected in the query image, and searching the building search DB for a building matching the extracted features. Therefore, building search performance can be improved.

Abstract: A data learning device in a deep learning network characterized by a high image resolution and a thin channel at an input stage and an output stage and a low image resolution and a thick channel in an intermediate deep layer includes a feature information extraction unit configured to extract global feature information considering an association between all elements of data when generating an initial estimate in the deep layer; a direct channel-to-image conversion unit configured to generate expanded data having the same resolution as a final output from the generated initial estimate of the global feature information or intermediate outputs sequentially generated in subsequent layers; and a comparison and learning unit configured to calculate a difference between the expanded data generated by the direct channel-to-image conversion unit and a prepared ground truth value and update network parameters such that the difference is decreased.

Abstract: A fast object detection method and a fast object detection apparatus using an artificial neural network. The fast object detection method includes obtaining an input image; inputting the obtained input image into an object detection neural network using a plurality of preset bounding boxes; and detecting an object included in the input image by acquiring output data of the object detection neural network.

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: A fast object detection method and a fast object detection apparatus using an artificial neural network. The fast object detection method includes obtaining an input image; inputting the obtained input image into an object detection neural network using a plurality of preset bounding boxes; and detecting an object included in the input image by acquiring output data of the object detection neural network.

Abstract: A machine learning method for learning how to form bounding boxes, performed by a machine learning apparatus, includes extracting learning images including a target object among a plurality of learning images included in a learning database, generating additional learning images in which the target object is rotated from the learning images including the target object, and updating the learning database using the additional learning images.

Abstract: An apparatus and method for searching for a building on the basis of an image and a method of constructing a building search database (DB) for image-based building search. The method includes constructing a building search DB, receiving a query image from a user terminal, detecting a region to which a building belongs in the query image, extracting features of the region detected in the query image, and searching the building search DB for a building matching the extracted features. Therefore, building search performance can be improved.

Abstract: A server for providing a city street search service includes a street information database configured to store city street images, a feature selection unit configured to select at least one feature according to a predetermined criterion when a city street image for searching and two or more features for the image are received from a user terminal, a candidate extraction unit configured to extract a candidate list of a city street image, a feature matching unit configured to match the city street image for registration included in the extracted candidate list and the at least one selected feature, and a search result provision unit configured to provide the user terminal with a result of the matching as result information regarding the city street image for searching.

Abstract: An omni-channel management method includes receiving a distribution channel information and a product purchase information from each of a plurality of distribution channels, and integrating the received distribution channel information and the product purchase information; applying an image identification technology and a voice identification technology to retrieve and integrate information related to a product; receiving and integrating product information from a manufacturer, a seller, or a web site; verifying consistency of the integrated product information by checking whether detailed information included in the integrated product information is related to a same product; and providing the consistency-secured product information to a consumer and a plurality of distribution channels.

Abstract: A machine learning method for learning how to form bounding boxes, performed by a machine learning apparatus, includes extracting learning images including a target object among a plurality of learning images included in a learning database, generating additional learning images in which the target object is rotated from the learning images including the target object, and updating the learning database using the additional learning images.

Abstract: An apparatus for food search service includes a food region extractor configured to perform detection in regions in an image where food is present and extract a plurality of candidate regions; a candidate region refiner configured to cluster the candidate regions into groups according to a ratio of overlap between the candidate regions; and a search result generator configured to determine a position of a food region and a food item from the grouped candidate regions.

Abstract: A method for extracting features from an image for use in a computing device, the method comprising: producing Gaussian Scale Space (GSS) images in the type of a pyramid from the image inputted to the computing device; performing a Scale Normalized Laplacian Filtering on the GSS images; detecting interest points from the images that are subject to the Scale Normalized Laplacian Filtering; and extracting features of the image using the detected interest points.

Abstract: An apparatus for food search service includes a food region extractor configured to perform detection in regions in an image where food is present and extract a plurality of candidate regions; a candidate region refiner configured to cluster the candidate regions into groups according to a ratio of overlap between the candidate regions; and a search result generator configured to determine a position of a food region and a food item from the grouped candidate regions.

Abstract: A server for providing a city street search service includes a street information database configured to store city street images, a feature selection unit configured to select at least one feature according to a predetermined criterion when a city street image for searching and two or more features for the image are received from a user terminal, a candidate extraction unit configured to extract a candidate list of a city street image, a feature matching unit configured to match the city street image for registration included in the extracted candidate list and the at least one selected feature, and a search result provision unit configured to provide the user terminal with a result of the matching as result information regarding the city street image for searching.

Abstract: Provided is a visual search system. The visual search system according to an embodiment of the inventive concept may include a database that stores characteristic information for a visual search, and a visual search update server that updates the DB based on an image of a target captured by a moving object or a fixed object. According to an embodiment of the inventive concept, it is possible to efficiently update a DB for a visual search because images captured by not only the moving object but also the fixed object are used.

Abstract: Provided are a shoe image retrieval apparatus and method using a matching pair, which can accurately retrieve image information corresponding to an inputted image from the database and provide the retrieved image information, by normalizing a correspondence relation in consideration of geometric image transformation about the matching pair and allowing a similar image to be retrieved by applying the normalized correspondence relation. It is possible to detect optimum geometric image transformation from a matching pair between the inputted shoe image and the image stored in the database and simultaneously retrieve a plurality of objects in the inputted shoe image based on the detected geometric image transformation, thereby providing an efficient shoe retrieval service.

Abstract: A method and system for extracting a visual descriptor using a feature selection are provided. The system includes an image input unit configured to receive an image, a candidate feature point group detecting unit configured to detect a point having a local maximum or minimum of local region filtering in scale-space images as being included in a candidate feature point group, a feature point selecting unit configured to calculate an importance for each candidate feature point, depending on its characteristics, select the candidate feature point as a feature point when its importance is greater than the predetermined threshold value, a dominant orientation calculating unit configured to calculate a dominant orientation of the selected feature point and a visual descriptor extracting unit configured to extract a patch for each feature point, according to its scale, location and dominant orientation, and extract a visual descriptor from the patch.

Abstract: A method for extracting features from an image for use in a computing device, the method comprising: producing Gaussian Scale Space (GSS) images in the type of a pyramid from the image inputted to the computing device; performing a Scale Normalized Laplacian Filtering on the GSS images; detecting interest points from the images that are subject to the Scale Normalized Laplacian Filtering; and extracting features of the image using the detected interest points.

Abstract: Provided are a shoe image retrieval apparatus and method using a matching pair, which can accurately retrieve image information corresponding to an inputted image from the database and provide the retrieved image information, by normalizing a correspondence relation in consideration of geometric image transformation about the matching pair and allowing a similar image to be retrieved by applying the normalized correspondence relation. It is possible to detect optimum geometric image transformation from a matching pair between the inputted shoe image and the image stored in the database and simultaneously retrieve a plurality of objects in the inputted shoe image based on the detected geometric image transformation, thereby providing an efficient shoe retrieval service.