Abstract: A method and apparatus for recognizing a sign language or a gesture by using a three-dimensional (3D) Euclidean distance matrix (EDM) are disclosed. The method includes a two-dimensional (2D) EDM generation step for generating a 2D EDM including information about distances between feature points of a body recognized in image information by a 2D EDM generator, a 3D EDM generation step for receiving the 2D EDM and generating a 3D EDM by using a first deep learning neural network trained with training data in which input data is a 2D EDM and correct answer data is a 3D EDM by a 3D EDM generator, and a recognition step for recognizing a sign language or a gesture based on the 3D EDM.

Abstract: Deep learning-based automatic gesture recognition method and system are provided. The training method according to an embodiment includes: extracting a plurality of contours from an input image; generating training data by normalizing pieces of contour information forming each of the contours; and training an AI model for gesture recognition by using the generated training data. Accordingly, robust and high-performance automatic gesture recognition can be performed, without being influenced by an environment and a condition even while using less training data.

Abstract: A method for preparing a SiC ingot includes loading a composition of a raw material comprising a carbon source and a silicon source into a reactor; placing a plurality of seed crystal on one side of the reactor spaced apart from the composition; and sublimating the composition such that the SiC ingot grows from the plurality of seed crystal. A flow factor of the composition may be 5 to 35.

Abstract: A method for producing an ingot includes loading a raw material comprising a raw material powder having a D50 of 80 ?m or more into a reactor (loading step), controlling the internal temperature of the reactor such that adjacent particles of the raw material powder are interconnected to form a necked raw material (necking step), and sublimating components of the raw material from the necked raw material to grow an ingot (ingot growth step).

Abstract: An apparatus for producing an ingot includes a crucible body having an opening and in which raw materials are accommodated, and a lid assembly located at the opening and having a portion fixed to the crucible body. The lid assembly includes a placement hole having open upper and lower ends, a frame member arranged along a periphery of the opening while surrounding a periphery of the placement hole, and a core member located in the placement hole and movable upward and downward with respect to the frame member.

Abstract: A method for preparing a seed crystal including a protective film includes preparing i) a first layer composition of a first binder resin and a first solvent and ii) a second layer composition of a second binder resin, a filler, and a second solvent, applying the first layer composition to the rear surface of a seed crystal to form a first coating layer on the rear surface of the seed crystal and drying the first coating layer to form a first layer on the rear surface of the seed crystal, and applying the second layer composition onto the first layer to form a second coating layer on the first layer, followed by heat treating to form a second layer on the first layer wherein the first layer and the second layer are sequentially disposed on the rear surface of the seed crystal, and wherein the first layer has a thickness corresponding to 30% or less of the distance from the bottom surface of the first layer to the top surface of the second layer.

Abstract: Deep learning-based automatic gesture recognition method and system are provided. The training method according to an embodiment includes: extracting a plurality of contours from an input image; generating training data by normalizing pieces of contour information forming each of the contours; and training an AI model for gesture recognition by using the generated training data. Accordingly, robust and high-performance automatic gesture recognition can be performed, without being influenced by an environment and a condition even while using less training data.