Abstract: A processor-implemented method of implementing an attention mechanism in a neural network includes obtaining key-value coupling data determined based on an operation between new key data determined using a first nonlinear transformation for key data of an attention layer, and value data of the attention layer corresponding to the key data; determining new query data by applying a second nonlinear transformation to query data corresponding to input data of the attention layer; and determining output data of the attention layer based on an operation between the new query data and the key-value coupling data.

Abstract: A processor-implemented method includes: estimating, from frame images of consecutive frames acquired from one or more sensors, short-term ego-motion information of the one or more sensors; estimating long-term ego-motion information of the one or more sensors from the frame images; determining attention information from the short-term ego-motion information and the long-term ego-motion information; and determining final long-term ego-motion information of a current frame, of the consecutive frames, based on the long-term ego-motion information and the attention information.

Abstract: A modeling method searches for a sequence matched to a user input using a fluid animation graph generated based on similarities among frames included in sequences included in the fluid animation graph and models a movement corresponding to the user input based on a result of the searching. Provided also is a corresponding apparatus and a method for preprocessing for such modeling.

Abstract: A method of modeling an object includes defining a state transition probability and a state of each of a plurality of particles forming the object; changing a state of a particle defined to be in a first state among the plurality of particles to a second state; applying a movement model to a particle defined to be in the second state among the plurality of particles; and changing a state of the particle defined to be in the second state to the first state based on the state transition probability.

Abstract: A method and apparatus for identifying a driving lane extracts a road marking from an input image, generate a multi-virtual lane using the road marking and a segmentation image obtained by segmenting the input image into objects included in the input image based on a semantic unit, and identifies a driving lane of a vehicle based on a relative location of the driving lane in the multi-virtual lane.

Abstract: A method and apparatus for displaying a virtual route estimates a position of a vehicle based on sensing data received asynchronously from sensors, and outputs a three-dimensional (3D) virtual route generated by registering a driving environment model corresponding to a driving environment of the vehicle and the position of the vehicle in map information.

Abstract: Provided is a method of modeling a target object, the method including obtaining information about the target object including an arrangement of particles including target particles, generating coarse particles by down-sampling the target particles, modeling a movement of the target object based on the coarse particles, and redefining the target particles based on a result of the modeling.

Abstract: Disclosed are methods and apparatuses for indicating a lane, the method including detecting a lane marking from an input image, acquiring a bird's eye view image by applying an inverse perspective mapping to the detected lane marking, extracting control points corresponding to the lane marking from the bird's eye view image, and indicating a lane by re-projecting the control points to a space of the input image through an inverse transformation.

Abstract: A localization method and apparatus are provided. The method comprises generating a synthetic image corresponding to a current time period based on a three-dimensional (3D) color map corresponding to an image captured at a time period prior to the current time period, and position information at the current time period time, and determining final position information at the current time period based on the synthetic image corresponding to the current time period and an image captured at the current time period.

Abstract: A vehicle localization method and a vehicle localization apparatus are disclosed. The vehicle localization method includes estimating an initial position of a vehicle based on data sensed by a position sensor, determining a driving lane of the vehicle based on a front-view image captured from the vehicle, correcting the initial position of the vehicle based on the driving lane, and determining a position of the vehicle in the driving lane based on geometry information of a lane boundary in the front-view image.

Abstract: Provided are a method and a corresponding apparatus for modeling a deformable body including particles that define a strain energy generated by an external force with respect to a deformable body including particles, and control a displacement of the deformable body based on a volume of the particles corresponding to the displacement of the deformable body, where the displacement is determined based on the strain energy.

Abstract: A method and corresponding apparatus to model objects includes detecting an overlapping area between first and second objects each comprising particles. The method and corresponding apparatus also calculate, in the overlapping area, an action force between the first and the second objects. The method and corresponding apparatus model the first object and the second object based on the action force.

Abstract: A modeling method based on particles, the method including generating coarse particles by down-sampling target particles corresponding to at least a portion of a target object, calculating a correcting value enabling the coarse particles to satisfy constraints of the target object based on physical attributes of the target particles, applying the correcting value to the target particles, and redefining the target particles in response to the target particles to which the correcting value is applied satisfying the constraints.

Abstract: A processor-implemented learning method for an image segmentation includes training first duplicate layers, as duplications of trained first layers of a pre-trained model, so that a second feature extracted from a target image by the trained first duplicate layers is matched to a first feature extracted from a training image by the trained first layers; regularizing the trained first duplicate layers so that a similarity between the first feature and a third feature extracted from the training image by the regularized first duplicate layers meets a threshold; and training second duplicate layers, as duplications of trained second layers of the pre-trained model, to be configured to segment the target image based on the regularized first duplicate layers, the trained second layer being configured to segment the training image.

Abstract: A particle-based modeling method and apparatus may include searching for a second particle neighboring a first particle in a vector field at a current time, and updating a position of each of the first particle and the second particle in each vector field over time, based on correcting a velocity of each of the first particle and the second particle based on a variation satisfying an incompressible constraint condition for each of the first particle and the second particle.

Abstract: A method and apparatus for displaying a virtual route estimates a position of a vehicle based on sensing data received asynchronously from sensors, and outputs a three-dimensional (3D) virtual route generated by registering a driving environment model corresponding to a driving environment of the vehicle and the position of the vehicle in map information.

Abstract: A method and apparatus for modeling smoke turbulence is disclosed. The method of modeling smoke turbulence includes detecting a turbulent area comprising smoke turbulence in a low-resolution image, and searching for a high-resolution patch matching the turbulent area in a memory that comprises high-resolution patches representing smoke turbulence extracted from high-resolution images. The method also models the smoke turbulence by synthesizing the retrieved patch to the low-resolution image.

Abstract: A method and apparatus for identifying a driving lane extracts a road marking from an input image, generate a multi-virtual lane using the road marking and a segmentation image obtained by segmenting the input image into objects included in the input image based on a semantic unit, and identifies a driving lane of a vehicle based on a relative location of the driving lane in the multi-virtual lane.

Abstract: A method of displaying caustics, the method includes determining intersection positions at which rays emitted from a light source pass through particles of a first object and meet a second object; applying caustic textures to the intersection positions; and rendering the first object using a caustic map based on a result of the applying caustic textures to the intersection positions.

Abstract: Provided is a method of modeling a target object, that may obtain depth information from an image in which the target object is represented in a form of particles, obtain distance information between adjacent particles in the image, and detect a silhouette of the target object based on the depth information and the distance information between the adjacent particles in the image.