Research on Detection and Identification of Dense Rebar Based on Lightweight Network

Target detection technology has been widely used,while it is less applied in portable equipment as it has certain requirements for devices.For instance,the inventory of rebar is still manually counted at present.In this paper,a lightweight network that adapts mobile devices is proposed to accomplish the task more intelligently and efficiently.Based on the existing method of detection and recognition of dense small objects,the research of rebar recognition was implemented.After designing the multi-resolution input model and training the data set of rebar,the efficiency of detection was improved significantly.Experiments prove that the method proposed has the advantages of higher detection degree,fewer model parameters,and shorter training time for rebar recognition.