Camouflaged targets are a type of nonsalient target with high foreground and background fusion and minimal target feature information,making target recognition extremely difficult.Most detection algorithms for camoufl...Camouflaged targets are a type of nonsalient target with high foreground and background fusion and minimal target feature information,making target recognition extremely difficult.Most detection algorithms for camouflaged targets use only the target’s single-band information,resulting in low detection accuracy and a high missed detection rate.We present a multimodal image fusion camouflaged target detection technique (MIF-YOLOv5) in this paper.First,we provide a multimodal image input to achieve pixel-level fusion of the camouflaged target’s optical and infrared images to improve the effective feature information of the camouflaged target.Second,a loss function is created,and the K-Means++clustering technique is used to optimize the target anchor frame in the dataset to increase camouflage personnel detection accuracy and robustness.Finally,a comprehensive detection index of camouflaged targets is proposed to compare the overall effectiveness of various approaches.More crucially,we create a multispectral camouflage target dataset to test the suggested technique.Experimental results show that the proposed method has the best comprehensive detection performance,with a detection accuracy of 96.5%,a recognition probability of92.5%,a parameter number increase of 1×10^(4),a theoretical calculation amount increase of 0.03 GFLOPs,and a comprehensive detection index of 0.85.The advantage of this method in terms of detection accuracy is also apparent in performance comparisons with other target algorithms.展开更多
Onboard visual object tracking in unmanned aerial vehicles(UAVs)has attractedmuch interest due to its versatility.Meanwhile,due to high precision,Siamese networks are becoming hot spots in visual object tracking.Howev...Onboard visual object tracking in unmanned aerial vehicles(UAVs)has attractedmuch interest due to its versatility.Meanwhile,due to high precision,Siamese networks are becoming hot spots in visual object tracking.However,most Siamese trackers fail to balance the tracking accuracy and time within onboard limited computational resources of UAVs.To meet the tracking precision and real-time requirements,this paper proposes a Siamese dense pixel-level network for UAV object tracking named SiamDPL.Specifically,the Siamese network extracts features of the search region and the template region through a parameter-shared backbone network,then performs correlationmatching to obtain the candidate regionwith high similarity.To improve the matching effect of template and search features,this paper designs a dense pixel-level feature fusion module to enhance the matching ability by pixel-wise correlation and enrich the feature diversity by dense connection.An attention module composed of self-attention and channel attention is introduced to learn global context information and selectively emphasize the target feature region in the spatial and channel dimensions.In addition,a target localization module is designed to improve target location accuracy.Compared with other advanced trackers,experiments on two public benchmarks,which are UAV123@10fps and UAV20L fromthe unmanned air vehicle123(UAV123)dataset,show that SiamDPL can achieve superior performance and low complexity with a running speed of 100.1 fps on NVIDIA TITAN RTX.展开更多
基金Project supported by the Shandong Provincial Natural Science Foundation of China(No.ZR2020MF015)the Aerospace Science and Technology Innovation Institute Stabilization Support Project(No.ZY0110020009)。
文摘Camouflaged targets are a type of nonsalient target with high foreground and background fusion and minimal target feature information,making target recognition extremely difficult.Most detection algorithms for camouflaged targets use only the target’s single-band information,resulting in low detection accuracy and a high missed detection rate.We present a multimodal image fusion camouflaged target detection technique (MIF-YOLOv5) in this paper.First,we provide a multimodal image input to achieve pixel-level fusion of the camouflaged target’s optical and infrared images to improve the effective feature information of the camouflaged target.Second,a loss function is created,and the K-Means++clustering technique is used to optimize the target anchor frame in the dataset to increase camouflage personnel detection accuracy and robustness.Finally,a comprehensive detection index of camouflaged targets is proposed to compare the overall effectiveness of various approaches.More crucially,we create a multispectral camouflage target dataset to test the suggested technique.Experimental results show that the proposed method has the best comprehensive detection performance,with a detection accuracy of 96.5%,a recognition probability of92.5%,a parameter number increase of 1×10^(4),a theoretical calculation amount increase of 0.03 GFLOPs,and a comprehensive detection index of 0.85.The advantage of this method in terms of detection accuracy is also apparent in performance comparisons with other target algorithms.
基金funded by the National Natural Science Foundation of China(Grant No.52072408),author Y.C.
文摘Onboard visual object tracking in unmanned aerial vehicles(UAVs)has attractedmuch interest due to its versatility.Meanwhile,due to high precision,Siamese networks are becoming hot spots in visual object tracking.However,most Siamese trackers fail to balance the tracking accuracy and time within onboard limited computational resources of UAVs.To meet the tracking precision and real-time requirements,this paper proposes a Siamese dense pixel-level network for UAV object tracking named SiamDPL.Specifically,the Siamese network extracts features of the search region and the template region through a parameter-shared backbone network,then performs correlationmatching to obtain the candidate regionwith high similarity.To improve the matching effect of template and search features,this paper designs a dense pixel-level feature fusion module to enhance the matching ability by pixel-wise correlation and enrich the feature diversity by dense connection.An attention module composed of self-attention and channel attention is introduced to learn global context information and selectively emphasize the target feature region in the spatial and channel dimensions.In addition,a target localization module is designed to improve target location accuracy.Compared with other advanced trackers,experiments on two public benchmarks,which are UAV123@10fps and UAV20L fromthe unmanned air vehicle123(UAV123)dataset,show that SiamDPL can achieve superior performance and low complexity with a running speed of 100.1 fps on NVIDIA TITAN RTX.
