METHOD AND APPARATUS OF RECOGNIZING OBJECT

According to an embodiment of the present invention, provided are an object recognizing method and an object recognizing apparatus. The object recognizing method comprises the steps of: generating a feature map by extracting a feature from an input image in a neural network; extracting a region of i...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
1. Verfasser: JEON, BA ROM
Format: Patent
Sprache:eng ; kor
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:According to an embodiment of the present invention, provided are an object recognizing method and an object recognizing apparatus. The object recognizing method comprises the steps of: generating a feature map by extracting a feature from an input image in a neural network; extracting a region of interest corresponding to at least one interest object from the input image in parallel with generation of the feature map, and determining the number of object candidate regions which are used for detecting the interest object based on the size of the region of interest; and recognizing the interest object from the region of interest based on the number of the object candidate regions in the neural network. 일 실시예에 따른 객체 인식 방법 및 장치는 신경망에서 입력 영상으로부터 특징을 추출하여 특징 맵을 생성하고, 특징 맵의 생성과 병렬적으로, 입력 영상으로부터 적어도 하나의 관심 객체에 대응하는 관심 영역을 추출하고, 관심 영역의 크기에 기초하여 관심 객체의 검출을 위해 이용되는 객체 후보 영역의 개수를 결정하며, 신경망에서 객체 후보 영역의 개수에 기초하여 관심 영역으로부터 관심 객체를 인식한다.