In the field of electrical equipment management, the accurate detection of text from nameplates is critical for effective information retrieval and maintenance. Traditional methods relying on manual detection are often time-consuming and error-prone. This paper presents a novel approach utilizing the DBNet (Differentiable Binarization Network) algorithm for automated text detection on electric equipment nameplates. DBNet employs an instance segmentation methodology, integrating differentiable binarization into the training process to produce robust binary maps, thereby enhancing text detection performance. Our experiments demonstrate that the DBNet model achieves superior accuracy and speed compared to traditional detection methods, effectively managing complex backgrounds and various text orientations. The results indicate that this approach significantly simplifies post-processing steps while maintaining high detection efficiency, making it a reliable solution for the challenges in the text detection of electric equipment nameplates.