Target Detection Based on SSD-Mobilenet Model
LIU Yan
ZHU Zhiyu
ZHANG Bing
Abstract：In order to shorten the training time of the model and speed up the rate of the convergence of the network during the target detection,this paper adopts a method combining convolutional neural network and transfer learning.The detection speed of the SSD network is fast and the mobilenet lightweight network takes up less memory. SSD-Mobilenet model combines the advantages of SSD network and Mobilenet network. Firstly,the SSD-Mobilenet model is pre-trained with the COCO dataset to obtain the pa?rameters and bottleneck description factors of the model. Then the Pets dataset is used to retrain the full connection layer of the net?work. Using the idea of transfer learning,the model can converge in a short time with small datasets. The experimental results show that the total training time is about 11 hours,and the detection accuracy reaches 74.5%. This shows that the method combined with SSD-Mobilenet model and transfer learning can shorten the training time of the model,accelerate the convergence speed of the mod?el and have a high detection accuracy.
Keywords：target detectionconvolutional neural networksSSD-Mobilenettransfer learning
Publication Date：2019-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 52-56 )