Abstract:A bridge disease identification approach based on an enhanced YOLO v3 algorithm is suggested to increase the accuracy of apparent disease detection of concrete bridges under complex backgrounds. First, the YOLO v3 network structure is enhanced to better accommodate the dense distribution and large variation of disease scale characteristics, and the detection layer incorporates the squeeze and excitation (SE) networks attention mechanism module and spatial pyramid pooling module to strengthen the semantic feature extraction ability. Secondly, CIoU with better localization ability is selected as the loss function for training. Finally, the K-means algorithm is used for anchor frame clustering on the bridge surface disease defects dataset. 1363 datasets containing exposed reinforcement, spalling, and water erosion damage of bridges are produced, and network training is done after manual labelling and data improvement in order to test the efficacy of the algorithm described in this paper. According to the trial results, the YOLO v3 model has enhanced more than the original model in terms of precision rate, recall rate, Average Precision (AP), and other indicators. Its overall mean Average Precision (mAP) value has also grown by 5.5%. With the RTX2080Ti graphics card, the detection frame rate increases to 84 Frames Per Second, enabling more precise and real-time bridge illness detection.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to improve the detection accuracy in bridge surface defect detection under complex backgrounds. Specifically, the paper makes improvements on the following problems existing in the existing YOLO v3 algorithm in bridge surface defect detection: 1. **Dense distribution of defect scales and large scale changes**: Defects on the bridge surface are often densely distributed and have large scale changes, which pose challenges to defect detection. 2. **Insufficient feature extraction ability**: Although the YOLO v3 algorithm uses a multi - scale prediction method, the extracted features have low robustness and are not suitable for bridge defect detection under complex backgrounds. 3. **Poor positioning effect**: The traditional Intersection over Union (IoU) loss function cannot provide effective adjustment gradients under complex backgrounds, resulting in a decrease in defect positioning accuracy. To address these problems, the paper proposes an enhanced YOLO v3 algorithm. The main improvement points include: 1. **Spatial Pyramid Pooling module (SPP)**: Introduce the SPP module to better handle defect features of different scales and improve the richness of feature representation. 2. **Feature extraction network with SENet embedded**: Embed the attention mechanism of SENet in front of the three detection layers of YOLO v3, and enhance the feature expression ability by recalibrating channel weights. 3. **Use a better positioning loss function (CIoU)**: Introduce the CIoU loss function, which takes into account the distance between the center points of the prediction box and the ground - truth box, the overlap ratio, and the aspect ratio, making the bounding box regression more stable. 4. **Generate anchor boxes by K - means clustering**: Perform K - means clustering on the self - built bridge defect data set to generate more suitable anchor boxes to adapt to defects of different scales. Through these improvements, the paper aims to improve the accuracy and real - time performance of bridge surface defect detection, especially the detection effect under complex backgrounds. Experimental results show that the enhanced YOLO v3 algorithm is superior to the original YOLO v3 algorithm in terms of precision rate, recall rate, and average precision (AP). The overall mean average precision (mAP) value has increased by 5.5%, and the detection frame rate has reached 84 frames per second.

Enhanced YOLO v3 for precise detection of apparent damage on bridges amidst complex backgrounds

Application Research of Bridge Damage Detection Based on the Improved Lightweight Convolutional Neural Network Model

BD-YOLOv8s: enhancing bridge defect detection with multidimensional attention and precision reconstruction

Research on Multi-Apparent Defects Detection of Concrete Bridges Based on YOLOR

Small target disease detection based on YOLOv5 framework for intelligent bridges

A real-time multi-defect automatic identification framework for concrete dams via improved YOLOv5 and knowledge distillation

Bridge Damage Detection using a Single-Stage Detector and Field Inspection Images

Superficial Defect Detection for Concrete Bridges Using YOLOv8 with Attention Mechanism and Deformation Convolution

Concrete bridge surface damage detection using a single‐stage detector

YOLOv5s-GTB: light-weighted and improved YOLOv5s for bridge crack detection

Quantitative Detection of Typical Bridge Surface Damages Based on Global Attention Mechanism and YOLOv7 Network

Road disease detection algorithm based on YOLOv5s-DSG

Research on Asphalt Pavement Disease Detection Based on Improved YOLOv5s

Damage Detection and Localization of Bridge Deck Pavement Based on Deep Learning

YOLO Based Bridge Surface Defect Detection Using Decoupled Prediction

A novel YOLOv8-GAM-Wise-IoU model for automated detection of bridge surface cracks

Multidamage Identification in High-Resolution Concrete Bridge Component Imagery Based on Deep Learning

Pavement Disease Detection through Improved YOLOv5s Neural Network

Lightweight object detection network for multi‐damage recognition of concrete bridges in complex environments

Bridge detection method for HSRRSIs based on YOLOv5 with a decoupled head

Computer Vision-Based Bridge Damage Detection Using Deep Convolutional Networks with Expectation Maximum Attention Module