Supervisor:
Southwest Ordnance Industry Bureau
Organizer:
Chongqing Ordnance Industry Society
Chongqing University of Technology
Chongqing University of Technology
Research on infrared ship target detection and tracking method based on improved PP YOLOE and ByteTrack algorithms
| DOI: | 10.11809/bqzbgcxb2024.11.037 |
| Keywords: | infrared ship; object detection tracking; PP YOLOE; task alignment learning; ByteTrack |
| Abstract: | A infrared ship target detection and tracking method based on improved PP YOLOE and ByteTrack algorithms is proposed to quickly and accurately detect ship targets from infrared images collected during drone reconnaissance, and to continuously track them.In response to the multi scale, small target, and occlusion issues in infrared ship target detection, corresponding improvements have been made based on the PP YOLOE algorithm by adopting Task Alignment Learning (TAL), adding multiple sampling paths, and strengthening the detection head; Subsequently, in response to the situation where occlusion often leads to tracking loss during the tracking process, based on the ByteTrack algorithm, a certain enhancement was made by combining Kalman filtering with the Hungarian algorithm and adding ReID features to calculate appearance similarity.The experimental results show that the proposed method has high detection accuracy, good tracking effect, and can meet the needs of practical tasks. |
| Published: | 2024-11-30 |