Abstract: An apparatus includes a modified image generator generating modified images by modifying each unlabeled image, a pre-trainer to generate a feature vector for each modified image by using an artificial neural network-based encoder and train the encoder based on the feature vector for each modified image, a pseudo-label generator to generate a feature vector for each unlabeled training image, cluster the training images based on the feature vector for each training image, and generate a pseudo-label for at least one training image among the training images based on the clustering result, and a further trainer to generate a predicted label by using the trained encoder and a classification model including a classifier to generate a predicted label for an image input to the trained encoder based on a feature vector, and train the classification model based on the pseudo-label and predicted label for the at least one training image.

Abstract: An apparatus includes a modified image generator generating modified images by modifying each unlabeled image, a pre-trainer to generate a feature vector for each modified image by using an artificial neural network-based encoder and train the encoder based on the feature vector for each modified image, a pseudo-label generator to generate a feature vector for each unlabeled training image, cluster the training images based on the feature vector for each training image, and generate a pseudo-label for at least one training image among the training images based on the clustering result, and a further trainer to generate a predicted label by using the trained encoder and a classification model including a classifier to generate a predicted label for an image input to the trained encoder based on a feature vector, and train the classification model based on the pseudo-label and predicted label for the at least one training image.

Abstract: Provided are an apparatus and a method of compositing vehicle periphery images. The method of compositing vehicle periphery images includes selecting an object of interest from a rear image, adjusting a view area of the rear image based on the object of interest, and compositing a left image, the rear image, and a right image by reflecting a composite reference line according to the view area.

Abstract: A vehicle driving assistance method is provided. The vehicle driving assistance method may comprise receiving an image of surroundings of a vehicle; receiving driving information of the vehicle; adjusting a region of interest (an ROI) in the image based on the driving information; detecting an object from the ROI; determining a probability of collision between the object and the vehicle; and outputting a signal based on the probability of collision.

Abstract: Provided is a method of lane change support. The method comprises step of detecting lane lines by analyzing an image obtained by a first image sensor provided on a side of a vehicle and step of setting a vehicle detection region for detecting a moving object in the image obtained by the first image sensor based on the detected lane lines and step of detecting the moving object in the set vehicle detection region and judging a possibility of collision between the vehicle and the detected object and step of providing lane change information indicating whether it is dangerous for the vehicle to change lanes based on the result of judging the possibility of collision.

Abstract: Provided are a collision avoidance method using a depth sensor, the method comprises receiving depth-based image information and identifying a path in the received depth-based image information, determining a depth level for each region of the depth-based image information, setting one or more distance-based sensing regions on the identified path based on the determined depth level, determining whether an object is detected in each of the set distance-based sensing regions and outputting a control signal for controlling the operation of a transport when determining that the object has been detected.