近年来,"无人机+"消防、"无人机+"巡检、"无人机+"应急等领域广泛应用无人机,对无人机遥感图像进行识别是在各领域实现"无人机+"的关键.但是,无人机遥感图像复杂的信息,大量的小目标给目标检测带来了巨大的挑战,随着深度学习时代的到来,在目标检测性能提升的同时,随之而来的是模型参数量的大量增加.基于此,以一种轻量化的YOLOv5s改进模型应用到无人机遥感图像识别当中,在提高目标检测性能的同时尽可能减少模型的参数量.实验结果表明,改进后的模型较原始的YOLOv5s模型mAP50和mAP50:95分别提高了7.9和4.3个百分点,模型参数量只增加0.14 M.
Application to UAV remote sensing image recognition with a lightweight improved model of YOLOv5s
In recent years,UAV have been widely used in various fields such as"UAV+"firefighting,"UAV+"inspection,"UAV+"emergency response,etc.The recognition of remote sensing images of UAVs is the key to realize"UAV+"in various fields.However,the complex information of UAV remote sensing images and a large number of small targets bring great challenges to target detection.With the arrival of the deep learning era,while the performance of target detection is improved,the number of model parameters increases substantially.Based on this,in this paper,a lightweight YOLOv5s improved model is applied to UAV remote sensing image recognition,which improves the target detection performance while reducing the number of model parameters as much as possible.The experimental results show that the improved model improves mAP50 and mAP50:95 by 7.9 and 4.3 percent-age point,respectively,compared with the original YOLOv5s model,and the number of model parameters increases by only 0.14 M.
万方数据
