Abstract: A method and a device for video moment retrieval and highlight detection may include obtaining an inverted token of a text query based on an original token of the text query, obtaining video feature vectors of video clips, and performing an interaction between the text query and video clips by simultaneously using the original token, the inverted token and the video feature vectors.

Abstract: A battery pack, in which an internal temperature and an external temperature of the battery pack are measured at a predetermined period from different temperature sensors installed in each of battery cells, a BMS, and a pack case of the battery pack and then be compared with a predetermined reference value to perform battery overheating diagnosis, in particular, determine whether causes of the overheating are due to internal or external factors of the battery pack, and a method for diagnosing the overheating.

Abstract: The present disclosure relates to a watch protection film fabrication system based on watch model profile data, which includes a modular server that fabricates WatchShells based on a dimension of a watch. The watch protection film fabrication system also includes a user terminal that purchases the WatchShells fabricated on the modular server.

Abstract: The present disclosure relates to a watch protection film fabrication system based on watch model profile data, which includes a modular server that fabricates WatchShells based on a dimension of a watch. The watch protection film fabrication system also includes a user terminal that purchases the WatchShells fabricated on the modular server.

Abstract: A method and an apparatus for detecting malfunction of LIDAR, and a method and an apparatus for generating data therefor are disclosed. According to an aspect of the present disclosure, there is provided a method for detecting malfunction of a LIDAR, including: acquiring point cloud data from the LIDAR; determining whether the number of points included in the point cloud data is less than a point count threshold; and determining whether the LIDAR has a fault based on the point cloud data by using a LIDAR malfunction detection model when the number of points is greater than or equal to the point count threshold, wherein the LIDAR malfunction detection model is a pre-trained model based on a normal dataset and at least one LIDAR-abnormal dataset, the normal dataset includes preset statistics and point cloud data acquired from the LIDAR in a normal state, and the LIDAR-abnormal dataset includes point cloud data generated based on the normal dataset.

Abstract: A robot safety monitoring system includes a stereo camera set configured to photograph a robot and a monitoring area around the robot and a processor connected to the stereo camera set. The stereo camera set includes a first camera including a plurality of first camera modules spaced apart from each other, and a second camera including a plurality of second camera modules spaced apart from each other. Based on a first direction, the first camera and the second camera have different fields of view. The processor is configured to set the monitoring area, determine whether an object is present in the monitoring area, and control the robot based on determining that an object is present in the monitoring area.

Abstract: The present invention relates to a display pixel circuit using ferroelectric thin-film transistors and a method of driving the same. More particularly, the present invention relates to a display pixel circuit in which a volatile switch function and a non-volatile memory function are selectively applied by using a plurality of ferroelectric thin-film transistors including a ferroelectric gate-insulating film having a high dielectric constant (high-k) and a method of driving the display pixel circuit.

Abstract: The present invention relates to a method for preparing a furan dicarboxylic acid compound, which includes preparing a hydroxyl furfural compound from a hydroxyl-free furfural compound; and preparing a furan dicarboxylic acid compound by an oxidation reaction of the hydroxyl furfural compound in the presence of a heterogeneous catalyst.

Abstract: A battery management system includes a battery monitoring integrated circuit (BMIC) that diagnoses a state of a battery, a first overvoltage detection unit that generates a first detection signal by comparing a battery voltage of the battery with a first reference voltage, a second overvoltage detection unit that generates a second detection signal by comparing the battery voltage with a second reference voltage different from the first reference voltage, and a micro controller unit (MCU) that controls the battery according to a diagnosis signal from the BMIC and diagnoses an abnormality of an apparatus according to the first and second detection signals, and a battery management method.

Abstract: Provided is a catalyst for preparing 1,3-cyclopentanediol by hydrogenation of 4-hydroxy-2-cyclopentenone, the catalyst including: a carrier including ?-Al2O3; and an active metal supported on the carrier.

Abstract: The present invention relates to a heterogeneous catalyst for making a dicarboxyl acid aromatic heterocyclic compound and a method for preparing a dicarboxyl acid aromatic heterocyclic compound, and according to the present invention, an oxide with improved yield and purity can be produced by an oxidation reaction of a bio-based aromatic heterocyclic compound under a heterogeneous catalyst.

Abstract: The present invention relates to a method for preparing a dicarboxyl acid aromatic heterocyclic compound and a dicarboxyl acid aromatic heterocyclic compound obtained thereby. According to the present invention, an oxide with improved yield and purity can be produced by an oxidation reaction of a bio-based hydroxyl-free aromatic heterocyclic compound under a heterogeneous catalyst.

Abstract: A refrigerator has a shelf seating protrusion and a separate shelf seating member. The shelf seating protrusion is not formed at an approximate mid-point, height wise, of a storage space defined by an inner casing. The shelf seating member may be coupled to the inner casing to support a shelf assembly. An arrangement of the shelf seating protrusions and the shelf seating member is configured to reduce deformation and cracking of the inner casing due to thermal impact in a manufacturing process or when items are stored on the shelf assembly.

Abstract: The present invention provides polyester and a method for preparing same, the polyester including a repeating unit derived from dihydroxyalkyl tetrahydrofuran obtained from biomass; and a repeating unit derived from short-branched aliphatic dicarboxylic acid. Thus, the polyester has a carbon reduction effect and may be utilized as a novel biomass polyethylene terephthalate (PET) for packaging, by having excellent transparency, impact resistance, flexibility and elastic modulus compared to conventional PET.

Abstract: Provided is a heterogeneous catalyst for preparing a dicarboxylic acid aromatic heterocyclic compound, the heterogeneous catalyst including an active metal; and a basic metal oxide carrier capable of carrying the active metal and forming a complex by coordinate bonding with the dicarboxylic acid aromatic heterocyclic compound.

Abstract: An artificial intelligence (AI) system capable of simulating functions of a human brain, such as recognition and judgment, by using the machine learning algorithm such as deep learning, and an application thereof are provided. A method of learning multi-modal data according to the AI system and an application thereof includes: obtaining first context information representing a characteristic of a first signal and second context information representing a characteristic of a second signal by using a first learning network model; obtaining hidden layer information based on the first context information and the second context information by using a second learning network model; obtaining a correlation value representing a relation degree between the hidden layer information by using the second learning network model; and learning the hidden layer information in which the correlation value is derived as a maximum value.

Abstract: A robot safety monitoring system includes a stereo camera set configured to photograph a robot and a monitoring area around the robot and a processor connected to the stereo camera set. The stereo camera set includes a first camera including a plurality of first camera modules spaced apart from each other, and a second camera including a plurality of second camera modules spaced apart from each other. Based on a first direction, the first camera and the second camera have different fields of view. The processor is configured to set the monitoring area, determine whether an object is present in the monitoring area, and control the robot based on determining that an object is present in the monitoring area.

Abstract: A refrigerator has a shelf seating protrusion and a separate shelf seating member. The shelf seating protrusion is not formed at an approximate mid-point, height wise, of a storage space defined by an inner casing. The shelf seating member may be coupled to the inner casing to support a shelf assembly. An arrangement of the shelf seating protrusions and the shelf seating member is configured to reduce deformation and cracking of the inner casing due to thermal impact in a manufacturing process or when items are stored on the shelf assembly.

Abstract: An artificial intelligence (AI) system capable of simulating functions of a human brain, such as recognition and judgment, by using the machine learning algorithm such as deep learning, and an application thereof are provided. A method of learning multi-modal data according to the AI system and an application thereof includes: obtaining first context information representing a characteristic of a first signal and second context information representing a characteristic of a second signal by using a first learning network model; obtaining hidden layer information based on the first context information and the second context information by using a second learning network model; obtaining a correlation value representing a relation degree between the hidden layer information by using the second learning network model; and learning the hidden layer information in which the correlation value is derived as a maximum value.

Abstract: The present invention relates to a method for measuring a fracture toughness using an instrumented indentation testing, which measures a load and an indentation depth in real time while applying a load to a specimen by an indenter having a flat punch shape.