Camouflage people detection via strong semantic dilation network

Fang, Zheng; Zhang, Xiongwei; Deng, Xiaotong; Cao, Tieyong; Zheng, Changyan

doi:10.1145/3321408.3326662

Cited by 14 publications

(15 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The channel information, spatial information, and multiscale contextual semantic information can enrich the semantic features and construct the strong semantic information of camouflage targets. Strong semantic information can establish the mapping between the shallow features and the target attributes, which is helpful in detecting camouflage targets when the shallow features are severely damaged 6 , 7 , 16 . The attention skip-connection based on the CBAM module is proposed in this paper.…”

Section: Methodsmentioning

confidence: 99%

“…In this paper, the ASPP module is constructed to process the feature maps generated by the backbone and capture the multiscale contextual semantic information of the camouflage targets. Due to the high similarities between the camouflage targets and the natural environment, CNN with a small receptive field pays more attention to local features, which easily leads to misjudgment between the target and the background 7 . The construction of the ASPP module can expand the receptive field of COD-YOLO and improve the ability of the algorithm to aggregate and represent the target semantic information.…”

Section: Methodsmentioning

confidence: 99%

“…At present, there are relatively few datasets for the detection of camouflage targets. A public dataset named camouflage people detection dataset (CPDD) has been constructed 6 , 7 and some camouflage target detection algorithms 9 – 12 have conducted experiments based on CPDD, which contains targets with different camouflage efficiency in the image. In order to verify the detection accuracy of the proposed COD-YOLO algorithm on targets with strong camouflage efficiency, a strong camouflage efficiency target detection dataset named SCETD is constructed based on CPDD in this paper.…”

Section: Strong Camouflage Efficiency Target Datasetmentioning

confidence: 99%

“…In order to verify the detection accuracy of COD-YOLO proposed in this paper on targets with strong camouflage efficiency, the SCETD is constructed. The remaining 2458 images of CPDD are selected to construct the SCETD dataset according to the following principles 7 .…”

Section: Strong Camouflage Efficiency Target Datasetmentioning

confidence: 99%

“…However, due to the “checkerboard effect” caused by deconvolution, the detection accuracy of this algorithm is limited. Fang et al 7 . built a strong semantic dilatation network (SSDN), in which series dilatation convolutions are introduced to expand the receptive field of the neural network to obtain more semantic information about the camouflage targets.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Camouflage target detection based on strong semantic information and feature fusion

Yan,

Hu,

et al. 2023

J. Electron. Imag.

View full text Add to dashboard Cite

Aiming at the detection difficulties in camouflage target detection, such as the high similarity between the target and its background, serious damage to the edge, and strong concealment of the target, a camouflage target detection algorithm YOLO of camouflage object detection based on strong semantic information and feature fusion is proposed. First, the attention mechanism convolutional block attention module (CBAM) is constructed to highlight the important channel features and target spatial locations to further aggregate rich semantic information from the high-level feature map. Then the atrous spatial pyramid pooling module is constructed to repeatedly sample the multiscale feature maps to expand the receptive field of the neural network, reduce feature sparsity in the process of convolution, and ensure dense features and multiscale contextual semantic information enter the feature fusion module. Finally, the attention skip-connections are constructed based on the CBAM module for fusing the original feature maps extracted by the backbone network to the corresponding detection outputs so as to eliminate the redundant features as well as enrich the target information of the network outputs. In order to fully verify the performance of the proposed algorithm, a camouflage target detection dataset named strong camouflage efficiency target dataset (SCETD) is constructed. Experimental results on SCETD show that the precision and recall of the proposed algorithm achieve 96.1% and 87.1%, respectively. The AP 0.5 and AP 0.5∶0.95 achieve 92.3% and 54.4%, respectively. The experimental results prove the effectiveness of the proposed method in detecting camouflage targets.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%