Deep learning based detection and counting algorithm for rebar images

Rebar as an important material in the construction process of water conservancy construction, the existing rebar counting method is time-consuming and labor-intensive for workers to count rebars. In order to achieve fast and accurate detection and counting of rebar, this work selects rebar images as the detection object and constructs a rebar detection and counting model based on the improved yolov5s algorithm. To address the problem where the model deployment is constrained by various hardware conditions such as memory space and processor computing power, a lightweight method based on ShuffleNet v2 is proposed in this paper. The attention mechanism of fused channels and spaces is further introduced to improve detection accuracy. Experimental results show that compared with the traditional models, the number of model parameters can be significantly reduced by 45.3% with acceptable detection accuracy sacrifice for proposed method.