公开(公告)号：US20240211749A1

公开(公告)日：2024-06-27

申请号：US18340996

申请日：2023-06-26

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Sujin JANG ,  Dae Ung JO

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: A method and apparatus with object estimation model training is provided. The method include generating a cross-correlation loss based on a first feature vector, generated using an interim first neural network (NN) model provided an input based on first input data about a target object, and a second feature vector generated using a trained second neural network provided another input based on second input data about the target object; and generating a trained first NN model, including training the interim first NN model based on the cross-correlation loss.

公开(公告)号：US20240153144A1

公开(公告)日：2024-05-09

申请号：US18331634

申请日：2023-06-08

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Dae Ung JO

IPC: G06T7/90 ,  G06T5/20 ,  G06T7/194 ,  G06V10/54 ,  G06V10/60

CPC classification number: G06T7/90 ,  G06T5/20 ,  G06T7/194 ,  G06V10/54 ,  G06V10/60 ,  G06T2207/10024

Abstract: A method and apparatus for composing a traffic light image are provided, where the method includes separating a foreground and a background of each of the one or more actual images using a color separator, training a color transformation matrix configured to estimate a color value of the foreground, training a position estimator configured to estimate a position of a center of the foreground from a brightness value of each of the one or more actual images, and generating a target image based on inputting an input image to the position estimator and the color separator.

公开(公告)号：US20250166393A1

公开(公告)日：2025-05-22

申请号：US18602299

申请日：2024-03-12

Applicant: SAMSUNG ELECTRONICS CO., LTD.

Inventor： Dae Ung JO ,  Jaewook YOO

IPC: G06V20/58 ,  G06V10/25 ,  G06V10/32 ,  G06V10/762

Abstract: Disclosed is a method of detecting a traffic light with an object recognition model configured to recognize traffic lights. The method includes: obtaining an input image from a camera included in a vehicle, the input image among frames, including previous frames, captured by the camera; estimating, based on prior information about traffic light objects, a first region of interest (RoI) for the input image; determining a second RoI based on the first RoI and based on detection results of the previous frames, wherein the detection of results correspond to recognition results of recognizing traffic lights in the previous frames by the object recognition model; and recognizing, by the object recognition model, a traffic light in the input image, wherein the recognizing is based on the input image and the second RoI.

公开(公告)号：US20250131707A1

公开(公告)日：2025-04-24

申请号：US18661277

申请日：2024-05-10

Applicant: Samsung Electronics Co., Ltd.

Inventor： Junho CHO ,  Moonsub BYEON ,  Dongwook LEE ,  Dae Ung JO

IPC: G06V10/94 ,  G06V10/764 ,  G06V10/774 ,  G06V10/82 ,  G06V20/58 ,  G06V20/70

Abstract: Provided are an object detection method and apparatus for an autonomous vehicle. A method of controlling a vehicle includes: detecting a surrounding environment using pieces of data on a driving environment of the vehicle and generating an indication of the surrounding environment; determining, among trained visual prompts received via a network from a server, a target visual prompt corresponding to the pieces of data; generating a merged image by combining a driving image of the autonomous vehicle with the target visual prompt using a predetermined operation; and performing object detection by inputting the merged image into a neural network model of the vehicle, the neural network model configured to infer objects from images inputted thereto.

公开(公告)号：US20250086469A1

公开(公告)日：2025-03-13

申请号：US18605119

申请日：2024-03-14

Applicant: Samsung Electronics Co., Ltd.

Inventor： Nayeon KIM ,  Sujin JANG ,  Dae Ung JO

IPC: G06N3/09

Abstract: A learning method of generating a vector map and a method and apparatus for generating a vector map are disclosed. The learning method includes converting a first feature extracted by inputting a first modality sensed by a first sensor to a student model into a first feature vector in a bird eye view (BEV) space, converting a second feature extracted by inputting a multi-modality including the first modality and a second modality sensed by a second sensor to a teacher model into a second feature vector in the BEV space, and learning the student model to generate a vector map corresponding to the first modality by back-propagating cross-correlation loss by dimension that causes the first feature vector to replicate a cross-correlation with the second feature vector to the student model.