Considerable road mileage puts tremendous pressure on pavement crack detection and maintenance. In practice, using a small parameter model for fast and accurate image-based crack segmentation is a challenge. However, current mainstream convolutional neural networks allocate computing resources to the same type of operators, which ignores the impact of different levels of feature extractors on the model performance. In this research, an end-to-end real-time pavement crack segmentation network (RIIAnet) is designed to improve performance by deploying different types of operators in separate layers of the network structure. Based on the extraction characteristics of cracks by convolution, involution, and asymmetric convolution, in the shallow layers the crack segmentation task is matched to extract rich low-level features by the designed asymmetric convolution enhancement module (ACE). Meanwhile, in the deep layers, the designed residual expanded involution module (REI) is used to enhance the high-level semantic features. Furthermore, the existing involution operator that fails to converge during training is improved. The ablation experiment demonstrates that the optimal ratio of the convolution and REI is 1/3 to obtain the optimal resource allocation and ACE improves the performance of the model. Especially compared with seven classical deep learning models of different structures, the results show that the proposed model reaches the highest MIOU, MPA, Recall, and F1 score of 0.7705, 0.9868, 0.8047, and 0.8485, respectively. More importantly, the parameter size of the proposed model is dramatically reduced, which is 0.04 times that of U-Net. In practice, the proposed model can be implemented in images with a high resolution of 2048 × 1024 in real time.

No abstract

Aiming at the difficulty of accurately segmenting pavement cracks in traditional detection methods, this paper proposes a lightweight real-time detection model named MFPNet with an end-to-end encoding and decoding structure. Firstly, in the encoding stage, based on the different extraction characteristics of the involution-G and convolution operators for cracks, the designed multi-type features parallel (MFP) module is used in the deep network to enhance the abstract semantic information with reducing information loss. Then, the simplified long connection structure is adopted in the decoding stage to maintain the detection speed without reducing the detection accuracy. Additionally, ablation experiments demonstrate the effectiveness of the designed module. What’s more, compared with other deep learning-based algorithms, the model proposed in this paper has excellent performance, and its MIOU, Recall, and F1 Score reach 0.7705, 0.8023, and 0.8485, respectively. In practice, MFPNet can be implemented in images with a high resolution of 2048×1024 in real time.

The defect of underground drainage pipes is the main inducing factor of urban disasters. However, existing detection robot has problems such as poor environmental adaptability and a low degree of automation for pipes. The deep learning-based amphibious robot designed in this study is a highly adaptable and efficient detection system. The designed ducted screw propelled wheels first provide power. Next, based on the multimodal sensors and the improved YOLOV4-Tiny, defect detection and 3D reconstruction are carried out. Finally, the defect location and image information are transmitted to the terminal for display by wire, and a detection report is generated. What’s more, the experimental results show that the MAP of the improved YOLOV4-Tiny in this research is improved by 2.18% compared with the baseline network, and the FPS is improved by 11.3 frames. The system proposed provides a new approach to drainage pipe inspection.