Apparatus for predicting movement of box-level object using only position information of box-level object
According to a preferred embodiment of the present invention, an apparatus for predicting movement of a box-level object using only box-level object position information receives only box-level object position information through a long short-term memory (LSTM) structure to predict a future object p...
Saved in:
Main Authors | , , , |
---|---|
Format | Patent |
Language | English Korean |
Published |
17.05.2022
|
Subjects | |
Online Access | Get full text |
Cover
Summary: | According to a preferred embodiment of the present invention, an apparatus for predicting movement of a box-level object using only box-level object position information receives only box-level object position information through a long short-term memory (LSTM) structure to predict a future object position at a box-level, thereby quickly predicting the movement of an object without using an optical flow requiring a large amount of calculation, and predicting the movement of the object even in a situation, in which the object to be predicted is hidden by other conditions, by predicting the position of the object only with the box-level object position information.
본 발명의 바람직한 실시예에 따른 박스-레벨 객체 위치 정보만을 이용한 박스-레벨 객체 이동 예측 장치는, 장단기 기억 신경망(long short term memory, LSTM) 구조를 통해 박스-레벨 객체 위치 정보만을 입력으로 받아 미래의 객체 위치를 박스-레벨로 예측함으로써, 큰 계산량을 필요로 하는 광학 흐름(optical flow)을 사용하지 않으면서도 객체의 움직임을 빠르게 예측할 수 있고, 또한, 박스-레벨 객체 위치 정보만을 이용하여 객체의 위치를 예측함으로써, 움직임을 예측할 객체가 다른 환경들에 의해 가려진 상황에서도 객체의 움직임을 예측할 수 있다. |
---|---|
Bibliography: | Application Number: KR20200149532 |