Detection and tracking of a moving object using an unmanned aerial vehicle (UAV) and evaluation of its effectiveness is currently one of the main tasks in the field of object detection. This article presents a hybrid method for detecting and tracking objects in UAV image sequences. The proposed method is based on the YOLOv3 model and the Kalman filter. An improved YOLO3 model using regular block and Mish activation function is used for the object detection part of the system. Afterwards, the Kalman filter is used to estimate the location of an object in the subsequent frames. During the experiments, the proposed hybrid method performed excellent results in terms of mean average precision (mAP) and center location error (CLE). The proposed model is implemented in the UAV object detection system using Raspberry Pi 4 model B for autonomous detection of moving objects by gimbal camera of the UAV.
A HYBRID METHOD FOR DETECTION AND TRACKING OF MOVING OBJECTS USING THE UAV
2023-06-28
CENTRAL ASIAN JOURNAL OF EDUCATION AND COMPUTER SCIENCES (CAJECS); Vol. 2 No. 3 (2023): CAJECS; 10-14 ; 2181-3213
Article (Journal)
Electronic Resource
English
DDC: | 629 |
Tracking moving objects using range data
Automotive engineering | 1997
|Tracking Moving Objects as Spatio-Temporal Boundary Detection
British Library Conference Proceedings | 2002
|MULTI SATELLITE DETECTION AND TRACKING OF MOVING OBJECTS
European Patent Office | 2020
|