Abstract: A method for training a language model for multimodal dialog includes generating a first token sequence by tokenizing one or more text messages included in multimodal dialog data, generating a second token sequence by inserting, based on an appearance position of an image included in the multimodal dialog data, one or more image presence tokens and one or more image non-presence tokens into the first token sequence, and replacing one or more tokens among a plurality of tokens generated by the tokenization with a mask token, generating one or more feature vectors for the image, and training an artificial neural network-based language model to predict, based on the second token sequence and the one or more feature vectors, an appearance position of the image in the multimodal dialog data.

Abstract: Disclosed is a text recognition method and apparatus. A text recognition post-processing method for reflecting user post-correction performed by a processor in an apparatus, the text recognition post-processing method includes training a deep learning post-processing model based on post-correction data comprising a partial image including post-correction target text and post-correction text when there is user post-correction for a text recognition result of an input image; and post-processing a text recognition result of another input image by applying the trained deep learning post-processing model.

Abstract: The disclosure relates to a data augmentation method for document classification based on artificial intelligence, which includes: obtaining a plurality of document data; measuring quality information of the plurality of document data; classifying the plurality of document data by quality using the measured quality information, and detecting a distribution of the plurality of document data classified by quality; and augmenting document data corresponding to a specific quality group based on the detected document data distribution by quality.

Abstract: The present disclosure relates to a document classification method and a document classification device. A document classification method performed by a processor inside a computing device according to an embodiment of the present disclosure may include: obtaining a predicted dimension which is a dimension of a feature vector necessary to classify a document image that has been input, through a learned dimension prediction model; generating a feature vector of the document image, through a feature extraction model based on the predicted dimension; and identifying a document type corresponding to the document image on the basis of the generated feature vector.

Abstract: The present disclosure relates to a method, an apparatus, a system, and a computer program for correcting table coordinate information. The present disclosure discloses a method for correcting table coordinate information performed by one or more processors in an apparatus, which may include: producing a list of first coordinate elements for a first axis or second coordinate elements for a second axis on the basis of coordinate information of a plurality of cells included in a table of a document image; performing grouping on the list of first coordinate elements or the list of second coordinate elements; and correcting the coordinate information of the cells of the table on the basis of the first coordinate elements or the second coordinate elements on which the grouping was performed.

Abstract: The present disclosure relates to a text-based document classification method and a document classification device. A text-based document classification method according to an embodiment of the present disclosure is performed by a processor inside a computing device, and may include: extracting, from a document image that has been input, words included in the document image; generating, based on a degree of similarity between the words, a word set including a configured number of words; generating a word set image by individually turning the word set into an image; extracting an important keyword used for document classification among words included in the word set image; and classifying a type of the document image by using the important keyword.

Abstract: The present disclosure relates to document quality-based image pre-processing that measures, for each item, the quality of an input document image that is to be analyzed and omits a part or the entirety of a pre-processing process, or performs an image pre-processing process to which an appropriate algorithm is applied, so that an unnecessary operation may be decreased and a processing time may be reduced, and a higher character recognition rate may be obtained than an image pre-processing process uniformly applied. A document quality-based image pre-processing method according to an embodiment of the present disclosure may be a document quality-based image pre-processing method performed by a processor in an apparatus, the method including measuring a document quality of an input document image, classifying the document quality, and performing image pre-processing corresponding to the document quality classified for the document image.

Abstract: The present disclosure relates to a method for displaying a document recognition result and a document recognition apparatus using same, wherein a method performed by a processor in a computing apparatus for displaying a document recognition result according to an embodiment of the present disclosure may include: extracting text from an input document image and matching multiple key values and values included in the text to generate key-value pairs; on the basis of a configured output item, extracting a key-value pair corresponding to the output item from the document image; and adding a highlight object to an area corresponding to the extracted key-value pair in the document image to generate a first image and displaying the first image.

Abstract: A method and system for providing an adaptive IP handoff triggering system for a wireless communication system that consists of either the same or heterogeneous wireless access technologies. A mobile node is notified of at least one pre-trigger timing parameter by a current access network. At least one variable handoff threshold is constructed with the mobile node based on the at least one pre-trigger timing parameter. An IP handoff is then initiated to a candidate access network when pilot beacon signal strength of the mobile node reaches the variable handoff threshold.

Abstract: The present invention provides a tunnel-based IP handoff process that can minimize the handoff latency associated with the standard Mobile IP registration. In the present invention, when neighboring nodes are detected, these detected nodes are identified as candidate nodes. Detection of just the existence of neighboring nodes should be possible even in situations where these neighboring nodes use different access technologies. For instance, pilot beacon signals may be taken as indication of the existence of candidate nodes although it may be impossible to compare them quantitatively if the beacon signals emanate from different access technologies. As soon as candidate nodes are identified, tunnels are established between the source node and the candidate nodes. The tunnels are used to forward data to the mobile node after a communication link between the mobile node and the source node goes down.

Abstract: The present invention provides a tunnel-based IP handoff process that can minimize the handoff latency associated with the standard Mobile IP registration. In the present invention, when neighboring nodes are detected, these detected nodes are identified as candidate nodes. Detection of just the existence of neighboring nodes should be possible even in situations where these neighboring nodes use different access technologies. For instance, pilot beacon signals may be taken as indication of the existence of candidate nodes although it may be impossible to compare them quantitatively if the beacon signals emanate from different access technologies. As soon as candidate nodes are identified, tunnels are established between the source node and the candidate nodes. The tunnels are used to forward data to the mobile node after a communication link between the mobile node and the source node goes down.

Abstract: A method and system for providing an adaptive IP handoff triggering system for a wireless communication system that consists of either the same or heterogeneous wireless access technologies. A mobile node is notified of at least one pre-trigger timing parameter by a current access network. At least one variable handoff threshold is constructed with the mobile node based on the at least one pre-trigger timing parameter. An IP handoff is then initiated to a candidate access network when pilot beacon signal strength of the mobile node reaches the variable handoff threshold.