A cascaded method for transmission tower number recognition in large scenes

摘要

Recognizing the transmission tower numbers is an import part of the automatic inspection of high-voltage transmissionlines. However, it's infeasible to accomplish this task effectively in one step giving the large scene images shot byunmanned aerial vehicles. In this paper, we present a cascaded framework consists of two CNN components: numberplate detection and serial number recognition. The proposed method reduces the difficulty of localizing numbercharacters in large scenes by leveraging the robust background, number plates. On the one hand, the proposed cascadedcoarse-to-fine method reduces the missing rate and improves the detection accuracy, on the other hand, the recognitioncomplexity is greatly reduced. The experimental results on our collected dataset demonstrate the effectiveness of theproposed method.