An Improved YOLOv10 Model with Ghost Mode for Camouflage Detection in Complex Environments

This paper focuses on the detection of camouflaged targets in complex environments. Given the complexity of combat scenarios and climates, this paper has developed a various weather, multi-scenario dataset for camouflage targets. To address the challenge of limited resources in complex environments, this paper proposes a lightweight network structure based on an improved Ghost mode in YOLOv10n. The main improvements are the CBS module and C2f module based on the Ghost convolution. Through comparative experiments with different networks, it validates the high accuracy of the proposed network in detecting camouflage target, while maintaining a low number of model parameters and computational requirements. The trained network is capable of effectively adapting to the constraints of limited resources in complex environments.