公开(公告)号：US20240078802A1

公开(公告)日：2024-03-07

申请号：US17903447

申请日：2022-09-06

Applicant: University of Electronic Science and Technology of China

Inventor： Fang HUANG ,  Shengyi Chen ,  Shuying Peng ,  Yinjie Chen

IPC: G06V10/94 ,  B64C39/02 ,  G06T7/20 ,  G06V20/17 ,  H04N5/232 ,  H04W4/029

CPC classification number: G06V10/95 ,  B64C39/024 ,  G06T7/20 ,  G06V20/17 ,  H04N5/23206 ,  H04N5/23245 ,  H04W4/029 ,  B64C2201/127 ,  G06T2207/10032

Abstract: A system for real-time target identification of an unmanned aerial vehicle (UAV) based on an embedded technique and an improved YOLO4 algorithm, and relates to the field of real-time target detection is disclosed. The system for real-time target identification of a UAV based on an embedded technique and an improved YOLO4 algorithm abandons a conventional “one-to-one” mode, and can be used for a plurality of UAV data processing applications, and multiple users in different geographical locations can further remotely manage and control the system. Particularly, in a target identification client of the present disclosure, to obtain a better real-time identification effect, a YOLOv4 target identification algorithm is improved, so that a better identification result and a faster identification speed are obtained.

公开(公告)号：US12080058B2

公开(公告)日：2024-09-03

申请号：US17903447

申请日：2022-09-06

Applicant: University of Electronic Science and Technology of China

Inventor： Fang Huang ,  Shengyi Chen ,  Shuying Peng ,  Yinjie Chen

IPC: G06T7/20 ,  B64C39/02 ,  G06V10/94 ,  G06V20/17 ,  H04N23/661 ,  H04N23/667 ,  H04W4/029 ,  B64U101/30

CPC classification number: G06V10/95 ,  B64C39/024 ,  G06T7/20 ,  G06V20/17 ,  H04N23/661 ,  H04N23/667 ,  H04W4/029 ,  B64U2101/30 ,  G06T2207/10032

Abstract: A system for real-time target identification of an unmanned aerial vehicle (UAV) based on an embedded technique and an improved YOLO4 algorithm, and relates to the field of real-time target detection is disclosed. The system for real-time target identification of a UAV based on an embedded technique and an improved YOLO4 algorithm abandons a conventional “one-to-one” mode, and can be used for a plurality of UAV data processing applications, and multiple users in different geographical locations can further remotely manage and control the system. Particularly, in a target identification client of the present disclosure, to obtain a better real-time identification effect, a YOLOv4 target identification algorithm is improved, so that a better identification result and a faster identification speed are obtained.