Container Damage Identification Based on RP-FCN

In this paper, a new upsampling algorithm is proposed to Fusion Upsampling, and a new RP-FCN network is presented based on ResNeXt50, Pyramid Pooling, Fusion Upsampling and FCN framework. The information perception system, operation control system and dispatching command system of the fully automated wharf are gradually perfected, but the automatic inspection module of the container has not been built yet. In this paper, the RP-FCN network is applied to the identification of container damage in port, in order to add a luster to the increasingly perfect fully automated terminal system. The experiment shows that the effect of the Fusion Upsampling is better than the traditional bilinear interpolation and deconvolution, and the effect of RP-FCN is better than that of FCN.