Abstract:Scene text detection is a fundamental research work in the field of image processing and has extensive application value. Segmentation-based methods have time-consuming feature processing, while post-processing algorithms are excellent. Real-time semantic segmentation methods use lightweight backbone networks for feature extraction and aggregation but lack effective post-processing methods. The pure convolutional network improves model performance by changing key components. Combining the advantages of three types of methods, we propose a Pure Convolutional Bilateral Segmentation Network (PCBSNet) for real-time natural scene text detection. First, we constructed a bilateral feature extraction backbone network to significantly improve detection speed. The low extraction detail branch captures spatial information, while the efficient semantic extraction branch accurately captures semantic features through a series of micro designs. Second, we built an efficient attention aggregation module to guide the efficient and adaptive aggregation of features from the two branches. The fused feature map undergoes feature enhancement to obtain more accurate and reliable feature representation. Finally, we used differentiable binarization post-processing to construct text instance boundaries. To evaluate the effectiveness of the proposed model, we compared it with mainstream lightweight models on three datasets: ICDAR2015, MSRA-TD500, and CTW1500. The F-measure scores were 82.9%, 82.8%, and 78.9%, respectively, and the FPS were 59.1, 94.3, and 75.5 frames per second. We also conducted extensive ablation experiments on the ICDAR2015 dataset to validate the rationality of the proposed improvements. The obtained results indicate that the proposed model significantly improves inference speed while enhancing accuracy and demonstrates good competitiveness compared to other advanced detection methods. However, when faced with curved text, the detection performance of PCBSNet needs to be improved.

Robust meter reading detection via differentiable binarization

A Coarse-to-fine Approach for Industrial Meter Detection and Its Application

Research on Digital Meter Reading Method of Inspection Robot Based on Deep Learning

Multi-Scale and Multi-Branch Convolutional Neural Network for Retinal Image Segmentation

Robust Pointer Meter Reading Recognition Method Under Image Corruption

Real-Time Scene Text Detection With Differentiable Binarization and Adaptive Scale Fusion

Reading Various Types of Pointer Meters under Extreme Motion Blur

A Mask RCNN based Automatic Reading Method for Pointer Meter

Automatic Meter Reading from UAV Inspection Photos in the Substation by Combining YOLOv5s and DeeplabV3

Research on reading recognition of pointer meter based on improved U-net network

An Automatic Reading Method for Pointer Meter Based on One-Stage Detector

An Improved YOLOv7-Based Model for Real-Time Meter Reading with PConv and Attention Mechanisms

Another Way: Direct Regression of Meter Readings for Circular Pointer Meter Images

PCBSNet: A Pure Convolutional Bilateral Segmentation Network for Real-Time Natural Scene Text Detection

A Robust Pointer Meter Reading Recognition Method Based on TransUNet and Perspective Transformation Correction

Towards Image-based Automatic Meter Reading in Unconstrained Scenarios: A Robust and Efficient Approach

Pointer Meter Reading Method Based on YOLOv8 and Improved LinkNet

Real-time detection algorithm for digital meters based on multi-scale feature fusion and GCS

Research on the Pointer Meter Reading Method Based on Adaptive Pooling and Hierarchical Fusion

Scale Value Guided Lite-FCOS for Pointer Meter Reading Recognition

Real Time Power Equipment Meter Recognition Based on Deep Learning