Abstract: The disclosure relates to a method of extracting image features based on a vision transformer, a method of performing embedding on an input image in units of patches and extracting visual features through global attention. An apparatus for extracting an image feature based on a vision transformer according to an embodiment of the disclosure includes a memory configured to store data and a processor configured to control the memory, wherein the processor is configured to perform embedding on multi-patches for an input image, extract feature maps for the embedding multi-patches, perform transformer encoding based on a neural network using the extracted feature maps, extract a feature of the input image through a final feature map extracted through the transformer encoding, and wherein the patches have different sizes.

Abstract: A scan driver includes a plurality of stage circuits. A last stage circuit among the plurality of stage circuits includes a self-initialization signal unit charging a node of the last stage circuit to a first level in response to a clock signal an initialization initializing the last stage circuit when the node is at the first level.

Abstract: Disclosed are a copper-clad laminate having a low dielectric constant, low dielectric loss, and high thermal conductivity properties, and a method for manufacturing the same, wherein an insulator substrate having a low dielectric constant, low dielectric loss and high thermal conductivity values is laminated onto a copper-foil layer and a portion of the copper-foil layer is patterned using selective etching such that a circuit pattern having a line width of 40 ?m or smaller is designed, such that the copper-clad laminate is used as a digital substrate.

Abstract: A display device includes: a display panel including pixels connected to scan lines; a gate driver including stages, a stage from among the stages being configured to: provide, as a carry signal, a carry clock signal of a carry clock line to a next stage from among the stages; and provide, as a scan signal, a scan clock signal of a scan clock line to a corresponding scan line from among the scan lines in response to a previous carry signal of a previous stage from among the stages; and a clock controller to connect the scan clock line to a power line in response to the carry clock signal, the power line being configured to provide a constant voltage.

Abstract: An emission driver is disclosed that includes stages. Each of the stages includes a first circuit part configured to apply a first node voltage to a first node, a second circuit part configured to apply a second node voltage to a second node and connected to the first circuit part through a third node, a third circuit part configured to apply a third node voltage of the third node or a voltage higher than the third node voltage to a fourth node based on the first node voltage and the second node voltage, and a fourth circuit part configured to generate an emission signal based on the first node voltage and a fourth node voltage of the fourth node.

Abstract: A scan driver includes a plurality of stage circuits. A last stage circuit among the plurality of stage circuits includes a self-initialization signal unit charging a node of the last stage circuit to a first level in response to a clock signal an initialization initializing the last stage circuit when the node is at the first level.

Abstract: A hydrogen storage system for detecting and restricting a pressure difference generated in the hydrogen storage system and a method for adjusting differential pressure therein are provided. The hydrogen storage system includes a controller that controls valves of a plurality of tanks. The controller measures pressure for each tank, measures a pipe pressure of pipe connected with the plurality of tanks, compares the pressure for each tank with the pipe pressure to determine whether differential pressure between the tank and the pipe is generated, determines whether it is expected to generate differential pressure between the tanks in an expected equilibrium temperature, when it is determined that the differential pressure between the tank and the pipe is not generated, and performs pressure equilibrium control between the tanks, when it is expected to generate the differential pressure between the tanks.

Abstract: An emission driver is disclosed that includes stages. Each of the stages includes a first circuit part configured to apply a first node voltage to a first node, a second circuit part configured to apply a second node voltage to a second node and connected to the first circuit part through a third node, a third circuit part configured to apply a third node voltage of the third node or a voltage higher than the third node voltage to a fourth node based on the first node voltage and the second node voltage, and a fourth circuit part configured to generate an emission signal based on the first node voltage and a fourth node voltage of the fourth node.

Abstract: Disclosed is a method for manufacturing a low dielectric constant and low dielectric loss material and part using inkjet printing. More specifically, disclosed is a method for manufacturing a low dielectric constant and low dielectric loss material and part by coating and heat-treating an ink composition containing fluoropolymer nanoparticles, ceramic nanoparticles, and polymer dispersant on a substrate.

Abstract: A light emitting display device including: a first pixel including a first lower storage electrode, a first gate electrode of a first driving transistor, and a first upper storage electrode; and a second pixel provided near the first pixel, and including a second lower storage electrode, a second gate electrode of a second driving transistor, and a second upper storage electrode. In a plan view, the first gate electrode and the second gate electrode have first sides facing each other, the first side of the first gate electrode is positioned inside a border of the first lower storage electrode or the first upper storage electrode in a plan view, and the first side of the second gate electrode is positioned inside a border of the second lower storage electrode or the second upper storage electrode in a plan view.

Abstract: 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.

Abstract: Disclosed is a method of manufacturing a stainless steel with excellent contact resistance for a polymer fuel cell separator. The method of manufacturing a stainless steel with excellent contact resistance for a polymer fuel cell separator according to an embodiment of the present disclosure includes: electrolyzing to remove a first passivation film formed on a cold-rolled thin sheet of a stainless steel comprising, in percent (%) by weight of the entire composition, C: greater than 0 to 0.1%, N: greater than 0 to 0.02%, Si: greater than 0 to 0.25%, Mn: greater than 0 to 0.2%, P: greater than 0 to 0.04%, S: greater than 0 to 0.02%, Cr: 22 to 34%, Ti: greater than 0 to 0.5%, Nb: greater than 0 to 0.5%, the remainder of iron (Fe) and other inevitable impurities; and immersing in a mixed acid solution of nitric acid and hydrofluoric acid to form a second passivation film on the stainless cold-rolled thin sheet.

Abstract: A display device includes a scan driver to supply scan signals to first and second scan lines, a data driver to supply a data signal to data lines, a sensor connected to sensing lines, and pixels including a light-emitting element, a driving transistor to control an amount of current supplied to the light-emitting element in response to a voltage of a first node, a switching transistor between a j-th data line and the first node, and including a gate electrode coupled to an i-th first scan line, and a sensing transistor coupled between a second node, which is between the light-emitting element and the driving transistor, and a k-th sensing line, and including a gate electrode coupled to an i-th second scan line, and wherein the sensor is to sense deterioration information of the light-emitting element in a state in which the switching and sensing transistors are turned on.

Abstract: Stainless steel for a fuel cell separator plate and a manufacturing method therefor are disclosed. The stainless steel for a fuel cell separator plate, according to one embodiment of the present invention, comprises: a stainless base material; and a passive film formed on the stainless base material, wherein a Cr/Fe atomic weight ratio in a 1 nm or less thickness region of the stainless base material, which is adjacent to an interface between the stainless and the passive film, is 0.45 or more. Therefore, by modifying the surface of the stainless steel for a fuel cell separator plate, a low interface contact resistance and a good corrosion resistance can be obtained, and a separate additional process such as precious metal coating can be removed, such that manufacturing costs are reduced and productivity can be improved.

Abstract: A scan driver for a display device may include a plurality of scan stage groups, each of the scan stage groups including a first scan stage and a second scan stage.

Abstract: A hydrogen storage system for detecting and restricting a pressure difference generated in the hydrogen storage system and a method for adjusting differential pressure therein are provided. The hydrogen storage system includes a controller that controls valves of a plurality of tanks. The controller measures pressure for each tank, measures a pipe pressure of pipe connected with the plurality of tanks, compares the pressure for each tank with the pipe pressure to determine whether differential pressure between the tank and the pipe is generated, determines whether it is expected to generate differential pressure between the tanks in an expected equilibrium temperature, when it is determined that the differential pressure between the tank and the pipe is not generated, and performs pressure equilibrium control between the tanks, when it is expected to generate the differential pressure between the tanks.

Abstract: A scan driver for a display device includes a plurality of scan stage groups, each of the scan stage groups including a first scan stage and a second scan stage.

Abstract: The disclosure relates to a method of extracting image features based on a vision transformer, a method of performing embedding on an input image in units of patches and extracting visual features through global attention. An apparatus for extracting an image feature based on a vision transformer according to an embodiment of the disclosure includes a memory configured to store data and a processor configured to control the memory, wherein the processor is configured to perform embedding on multi-patches for an input image, extract feature maps for the embedding multi-patches, perform transformer encoding based on a neural network using the extracted feature maps, extract a feature of the input image through a final feature map extracted through the transformer encoding, and wherein the patches have different sizes.

Abstract: 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.

Abstract: A scan driver for a display device includes a plurality of scan stage groups, each of the scan stage groups including a first scan stage and a second scan stage.