Improved single shot multibox detector target detection method based on deep feature fusion

Abstract

The feature layers of different layers in the single shot multibox detector (SSD) are independently used as the input of the classification network, so it is easy to detect the same object. This article proposes an improved SSD model based on deep feature fusion. In the SSD algorithm, the deep feature fusion between the target detection layer and its adjacent feature layer is used, including convolution kernels and pooling kernels of different sizes, down-sampling of low-level features and up-sampling of deconvolution of high-level features. The network is improved by combining the target frame recommendation strategy in the SSD algorithm and the frame regression algorithm. The experimental results show that the improved SSD algorithm improves the detection accuracy and detection rate of the target, and the effect is more obvious for the relatively small-scale target.

Intelligent Detection of Steel Defects Based on Improved Split Attention Networks

“…Feature Pyramid Networks (FPN) was proposed by Lin Tsung-Yi and others in 2017 ( Lin et al, 2017 ). FPN introduces multi-scale in the feature pyramid network and improves on the basis of the SSD multi-layer branching method ( Bai et al, 2021 ; Cui et al, 2021 ). Similar to the TDM (Top-Down Modulation) method, FPN is a top-down feature fusion method.…”

Attitude Stabilization Control of Autonomous Underwater Vehicle Based on Decoupling Algorithm and PSO-ADRC

“…The Active Disturbance Rejection Control (ADRC) algorithm studied in this paper ( Han, 2002 ; Ma et al, 2020 ; Sun et al, 2021 ; Tao et al, 2022b ), which inherits the characteristics of the PID algorithm, does not depend on the accurate model of the control object, while using modern signal processing techniques to improve its control process, unifies external disturbances and system errors as total disturbances, and then estimates and compensates them, which has the advantage of being easy to use in engineering and at the same time has a strong anti-disturbance capability. Depending on the Particle Swarm Optimization (PSO) ( Wang et al, 2013 ; Liu X. et al, 2021 ; Bai et al, 2022 ; Liu et al, 2022b ) which is characterized by fast convergence and high computational efficiency ( Li et al, 2019b ; Jiang et al, 2021b ; Liu et al, 2021c ; Huang et al, 2021 . ), Optimizing the parameters of ADRC using PSO can achieve better control effect.…”

Genetic Algorithm-Based Trajectory Optimization for Digital Twin Robots

“…These algorithms are used in different situations due to different computational principles ( Li et al, 2019a ; Jiang et al, 2021a ). Most of the current research on mobile robots uses improved algorithms to reduce their optimal path length or the number of iterations, and there is a lack of research on error control during the actual motion of the robot ( Li et al, 2019b ; Bai et al, 2021 ). In practice, the robot cannot follow the planned path to the target position during the movement due to the influence of environmental factors or the error in the coordination of various parts of the robot ( Chen et al, 2021b ; Chen et al, 2021c ).…”