Image-based intelligent detection of road cracks with high accuracy and efficiency is vital to the overall condition assessment of the pavement. However, significant problems of continuous cracks interruption and background discrete noise misidentification are frequently observed in current semantic segmentation of pavement cracks, which mainly caused by traditional segmentation convolutional neural networks. This paper proposes a skip-level round-trip sampling block structure with the implementation of convolutional neural networks, thereby constructed a novel pixel level semantic segmentation network called CrackW-Net. After that, two datasets, including the widely recognized Crack500 dataset and a self-built dataset, were used to train two versions CrackW-Net, FCN, U-Net and ResU-Net. Meanwhile, comparative experiments are conducted among all these network models for crack detection. Results show that CrackW-Net without residual block performs the best in the task of pavement crack segmentation.

Pixel-level crack detection is crucial in pavement performance assessment. Current deep learning-based detection methods first encode input images by multi-scale feature maps, then decode them to the output that has the same size as input. This process will lose detailed crack information. To tackle this problem, this paper proposed a novel network architecture, Enhanced High-Resolution Semantic Network (EHRS-Net), to maintain and enhance detailed information of the feature maps through convolution procedure, thus, improving the overall crack detection accuracy. The contributions of this paper are: (1) Proposed Resolution Maintain Flow (RMF), which is featured by three different semantic representation extraction flows in parallel with semantic information exchange; (2) Proposed Stacked Atrous Spatial Pyramid Pooling (SASPP) module to enhance the output of the semantic features; (3) Developed a new hybrid loss function to fit proposed network architecture. The proposed methods are evaluated on two pavement crack datasets: an expanded public crack forest dataset (CFD-ex) and a new dataset called HRSD (high-resolution semantic dataset). Comprehensive comparative experiments proved the superiority of the proposed method for pavement crack detection (93.353 % mPA (mean pixel accuracy) and 78.328% mIoU (mean intersection over union) on CFD-ex; 77.159% mIoU on HRSD), especially for tiny cracks and noised pavement cracks.

In order to detect pavement cracking with high accuracy, a type of three-dimensional (3D) pavement detection approach is proposed based on a two-dimensional (2D) empirical mode decomposition (EMD) algorithm, a nonparametric and data-driven sifting method which is able to detect pavement cracking without complicated convolution processes. By extending 2D EMD to the analysis of 3D pavement data, the proposed approach first decomposes pavement cracking data into subelements called intrinsic model functions (IMFs), which are monocomponent functions that have well-defined instantaneous frequencies. By combining double-phase standard deviation with structural element–based morphological algorithms, random noise is eliminated from the original collected 3D pavement data. The main cracking then is extracted using the EMD method, followed by binarization of the pavement distress images. Finally, pavement cracking is extracted by combining the region growing method with the morphology method on the binarized images. Experiments on different types of pavement cracking demonstrate the proposed algorithm to be accurate and superior to manual detection in terms of precision-recall curves. Moreover, the proposed approach gives a novel prospective for the measurement of pavement performance criteria.

Immediate and accurate detection of foreign metal objects (FMOs) in clothing products is important for guaranteeing human safety. This article proposes an online detection approach based on deep learning, which is suitable for detecting small FMOs from X-ray images of clothing packages. A conveyor belt X-ray scanning system is developed for image collection. The X-ray images are preprocessed by using the morphological erosion operation to improve the accuracy of FMOs detection. These images are then down-sampled to reduce the computation cost. Feature pyramid network (FPN) is adopted for aggregating feature maps with different resolutions, which proved to be effective for small FMOs detection. The stochastic gradient descent (SGD) is used to optimize a multitask loss. The trained model was tested offline on 200 X-ray images, which achieved precision = 0.999, recall = 0.988, F1-score = 0.993, and AP = 0.946. Compared to original Faster region-based convolutional neural networks (R-CNN), the proposed method significantly improved the performance for small FMOs detection in terms of precision and recall rate.