Real-Time Target Detection Method Based on Lightweight Convolutional Neural Network

Abstract

The continuous development of deep learning improves target detection technology day by day. The current research focuses on improving the accuracy of target detection technology, resulting in the target detection model being too large. The number of parameters and detection speed of the target detection model are very important for the practical application of target detection technology in embedded systems. This article proposed a real-time target detection method based on a lightweight convolutional neural network to reduce the number of model parameters and improve the detection speed. In this article, the depthwise separable residual module is constructed by combining depthwise separable convolution and non–bottleneck-free residual module, and the depthwise separable residual module and depthwise separable convolution structure are used to replace the VGG backbone network in the SSD network for feature extraction of the target detection model to reduce parameter quantity and improve detection speed. At the same time, the convolution kernels of 1 × 3 and 3 × 1 are used to replace the standard convolution of 3 × 3 by adding the convolution kernels of 1 × 3 and 3 × 1, respectively, to obtain multiple detection feature graphs corresponding to SSD, and the real-time target detection model based on a lightweight convolutional neural network is established by integrating the information of multiple detection feature graphs. This article used the self-built target detection dataset in complex scenes for comparative experiments; the experimental results verify the effectiveness and superiority of the proposed method. The model is tested on video to verify the real-time performance of the model, and the model is deployed on the Android platform to verify the scalability of the model.

Multi-network collaborative lift-drag ratio prediction and airfoil optimization based on residual network and generative adversarial network

“…As a conventional deep learning algorithm, CNN has good prediction performance Yun et al (2022) ; Sun et al (2022) . Based on the network architecture presented in Figure 4 , it is evident that the predictive ability of the neural network is affected by the number of single-layer convolution channels and the number of layers.…”

Large scale instance segmentation of outdoor environment based on improved YOLACT

“…In consideration of the fact that ResNet does not make full use of contextual information, this method ameliorates the acquisition of global information by pooling and fusion of multiple branches through a pyramid pooling module similar to that in PSPNet . [75][76][77][78][79] The extracted feature map is pooled several times, convolved, and upsampled to obtain a composite feature map with multiple scales consistent with the channels of the input feature map size, whose structure is shown in Figure 3, where the size of the kernel of Conv is 1 × 1 and activation function is ReLU.…”

“…In consideration of the fact that ResNet does not make full use of contextual information, this method ameliorates the acquisition of global information by pooling and fusion of multiple branches through a pyramid pooling module similar to that in PSPNet . 75–79 …”

A real-time and accurate detection approach for bucket teeth falling off based on improved YOLOX

“…Deep separable convolution (Buiu et al, 2020;Yun et al, 2022) is the core of the lightweight network MobileNet, which can achieve the same feature extraction effect as traditional convolution. Replacing the traditional convolution in PANet (path aggregation network) with the deep separable convolution can improve the problem that the running speed cannot be improved due to the high amount of computation in the traditional convolution, and at the same time reduce the model volume and network parameters.…”