| 摘要: |
| 针对无人机对于不同尺度的目标识别存在误检和漏检等问题，提出了一种多尺度融合机制的YOLOv8-FDT无人机目标识别算法。首先，在基线模型Neck层中添加动态上采样模块，旨在降低模型参数量，提高模型对于目标识别的实时性。此外，为了使得整个算法模型在特征融合阶段能够捕捉目标的不同尺度语义信息，融合自适应下采样和深度卷积，设计特征聚焦扩散金字塔网络(Feature Diffusion Pyramid Network，FDPN。通过无人机航拍数据集VisDrone2019的实验表明，改进后模型所有类别平均精度均值(mean Average Precision，mAP相较基线模型提升6.24%。 |
| 关键词: 无人机 目标识别 特征聚焦 多尺度融合 |
| DOI：10.20079/j.issn.1001-893x.240527001 |
|
| 基金项目:国家自然科学基金资助项目（62003151 |
|
| UAV Target Recognition Based on Multi-scale Features |
| ZHANG Bowen,XUE Bo |
| (School of Electrical and Information Engineering,Jiangsu Institute of Technology,Changzhou 213000，China) |
| Abstract: |
| In view of the problems of false detection and missed detection when an unmanned aerial vehicle（UAVdetects targets at different scales,a YOLOv8-FDT UAV algorithm model with a multi-scale fusion mechanism is proposed.First,a dynamic upsampling module is added to the Neck layer of the baseline model to reduce the number of model parameters and improve the real-time performance of the model for target recognition.In addition,in order to enable the entire algorithm model to capture different scale semantic information of the target in the feature fusion stage,adaptive downsampling and depth convolution are integrated to design the feature diffusion pyramid network(FDPN.Finally,experiments on the UAV aerial photography dataset VisDrone2019 show that the mean average precision(mAP of all categories of the improved model is increased by 6.24% compared with that of the baseline model. |
| Key words: UAV small target recognition focus feature multi-scale fusion |
