联合自注意力和分支采样的无人机图像目标检测

Joint self-attention and branch sampling for object detection on drone imagery

无人机图像目标检测在诸多领域被广泛应用,但受制于图像背景复杂、目标密集,目标尺度变化剧烈,现有的无人机图像目标检测算法检测效果不够精准。为解决此类问题,提出了一种联合自注意力和分支采样的无人机图像目标检测方法。首先,设计了自注意力和卷积相融合的嵌套残差结构以实现全局信息和局部信息的有效结合,让模型聚焦于待测目标,从而淡化复杂背景的影响。其次,设计了一种基于分支采样的特征融合模块以弥补目标信息丢失。最后,引入浅层细粒度特征图,新增了针对微小目标的改进检测头以缓解尺度剧烈变化,并基于此提出一种特征增强模块,用于捕获更多具有鉴别性的小目标特征。经实验验证,本文所提算法在多种场景中性能良好。其中s模型在VisDrone2019数据集上的mAP50和mAP分别达到59.3%和37.1%,相较于基线模型增长了5.6%和5.4%,在UAVDT数据集上的mAP50和mAP分别达到44.1%和24.9%,相较于基线模型提高了5.8%和3.2%。

Object detection on drone imagery is widely used in many fields.However,due to the complex⁃ity of the image background,the dense small objects and the dramatic scale changes,the existing object de⁃tection on drone imagery methods are not accurate enough.In order to solve this problem,we propose an accurate object detection method for drone imagery joint self attention and branch sampling.Firstly,a nested residual structure integrating self attention and convolution is designed to achieve the effective com⁃bination of global and local information,which makes the model to focus on the object area and ignore in⁃valid features.Secondly,we design a feature fusion module based on branch sampling to mitigate the loss of object information.Finally,an improved detector for small objects is added to alleviate the problem of sharp scale changes.Furthermore,we propose a feature enhancement module to obtain more discrimina⁃tive small object features.The experimental results show that the proposed algorithm performs well in var⁃ious scenarios.Specifically,the mAP50 and mAP of the s model on the VisDrone2019 reached 59.3%and 37.1%respectively,an increase of 5.6%and 5.4%compared with the baseline.The mAP50 and mAP on the UAVDT reached 44.1%and 24.9%respectively,an increase of 5.8%and 3.2%compared with the baseline.