Vehicle detection with category inference on video sequence data is an important but challenging task for urban traffic surveillance. The difficulty of this task lies in the fact that it requires accurate localization of relatively small vehicles in complex scenes and expects real-time detection. In this paper, we present a vehicle detection framework that improves the performance of the conventional Single Shot MultiBox Detector (SSD), which effectively detects different types of vehicles in real-time. Our approach, which proposes the use of different feature extractors for localization and classification tasks in a single network, and to enhance these two feature extractors through deconvolution (D) and pooling (P) between layers in the feature pyramid, is denoted as DP-SSD. In addition, we extend the scope of the default box by adjusting its scale so that smaller default boxes can be exploited to guide DP-SSD training. Experimental results on the UA-DETRAC and KITTI datasets demonstrate that DP-SSD can achieve efficient vehicle detection for real-world traffic surveillance data in real-time. For the UA-DETRAC test set trained with UA-DETRAC trainval set, DP-SSD with the input size of 300 × 300 achieves 75.43% mAP (mean average precision) at the speed of 50.47 FPS (frames per second), and the framework with a 512 × 512 sized input reaches 77.94% mAP at 25.12 FPS using an NVIDIA GeForce GTX 1080Ti GPU. The DP-SSD shows comparable accuracy, which is better than those of the compared state-of-the-art models, except for YOLOv3.

To address the challenges of coal fracture image recognition, including interference from gangue and multiscale fractures, a multiscale coal fracture segmentation network model to significantly enhance the recognition of coal fracture structures is proposed. The model significantly enhances the recognition of fracture structures based on a U-shaped network architecture and the incorporation of several advanced techniques, including transfer learning, depthwise separable atrous convolutions, and residual modules. Transfer learning, by leveraging pretrained visual geometry group 16-layer network weights, bolsters the feature extraction capabilities of an encoder. Simultaneously, the integration of depthwise separable atrous convolutions and residual modules optimizes a decoder, thereby improving segmentation accuracy and the robust recognition of fractures within images. Experimental results based on qualitative and quantitative data showed that the proposed model surpassed traditional convolutional neural networks, demonstrating proficiency in identifying multiscale fractures in complex coal images. The model was applied to the identification of fractures in roadway surrounding rock boreholes. By extracting fractures from borehole imaging videos and planar diagrams, and conducting cross-validation, the study precisely delineated the fracture distribution. Additionally, to improve coal seam gas extraction efficiency, the grouting and sealing range for cross-layer extraction boreholes was determined. Document Type: Original article Cited as: Wang, D., Li, L., Zhang, H., Li, S., Zhang, F., Xia, Y. Intelligent identification of coal fractures using an improved U-shaped network. Advances in Geo-Energy Research, 2025, 15(2): 129-141. https://doi.org/10.46690/ager.2025.02.05 References Alshammari, A. Construction of VGG16 convolution neural network (VGG16_CNN) classifier with NestNet-based segmentation paradigm for brain metastasis classification. Sensors, 2022, 22(20): 8076. Bai, C., Liu, X. The edge detection technology of CT image for study the growth of rock crack. Paper Presented at 2009 ISECS International Colloquium on Computing, Communication, Control, and Management, Sanya, China, 8-9 August, 2009. Chaurasia, A., Culurciello, E. Linknet: exploiting encoder representations for efficient semantic segmentation. Paper Presented at 2017 IEEE Visual Communications and Image Processing (VCIP), St. Petersburg, FL, USA, 10-13 December, 2017. Chen, L., Papandreou, G., Kokkinos, L., et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834-848. De Castro, A. R., Agnaou, M., Ahmadi-Sénichault, A., et al. Numerical porosimetry: Evaluation and comparison of yield stress fluids method, mercury intrusion porosimetry and pore network modelling approaches. Computers & Chemical Engineering, 2020, 133: 106662. Dai, Y., Li, C., Su, X., et al. Multi-scale depthwise separable convolution for semantic segmentation in street-road scenes. Remote Sensing, 2023, 15(10): 2649. Fang, H., Sang, S., Liu, S., et al. Methodology of three-dimensional visualization and quantitative characterization of nanopores in coal by using FIB-SEM and its application with anthracite in Qinshui basin. Journal of Petroleum Science and Engineering, 2019, 182: 106285. Hao, D., Tu, S., Zhang, C., et al. Quantitative characterization and three-dimensional reconstruction of bituminous coal fracture development under rock mechanics testing. Fuel, 2020, 267: 117280. He, K., Zhang, X., Ren, S., et al. Deep Residual Learning for Image Recognition. Paper Presented at 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27-30 June, 2016. He, M., Wang, Q. Rock dynamics in deep mining. International Journal of Mining Science and Technology, 2023, 33(9): 1065-1082. Jain, P., Dubey, A., Saba, L., et al. Attention-based UNet deep learning model for plaque segmentation in carotid ultrasound for stroke risk stratification: An artificial intelligence paradigm. Journal of Cardiovascular Development and Disease, 2022, 9(10): 326. Leng, B., Yang, H., Hou, G., et al. Rock mass trace line identification incorporated with grouping algorithm at tunnel faces. Tunnelling and Underground Space Technology, 2021, 110: 103810. Liu, S., Li, X., Wang, D., et al. Experimental study on temperature response of different ranks of coal to liquid nitrogen soaking. Natural Resources Research, 2021, 32: 1467-1480. Liu, Z., Liu, D., Cai, Y., et al. Application of nuclear magnetic resonance (NMR) in coalbed methane and shale reservoirs: A review. International Journal of Coal Geology, 2020, 218: 103261. Li, Y., Cui, H., Zhang, P., et al. Three-dimensional visualization and quantitative characterization of coal fracture dynamic evolution under uniaxial and triaxial compression based on CT scanning. Fuel, 2020, 262: 116568. Li, S., Tang, D., Xu, H., et al. Advanced characterization of physical properties of coals with different coal structures by nuclear magnetic resonance and X-ray computed tomography. Computers & Geosciences, 2012, 48: 220- 227. Liu, F., Wang, L. UNet-based model for crack detection integrating visual explanations. Construction and Building Materials, 2022, 322: 126265. Liu, J., Jiang, Y., Zhao, Y., et al. Crack edge detection of coal CT images based on LS-SVM. Paper Presented at 2009 International Conference on Machine Learning and Cybernetics, Hebei, China, 12-15 July, 2009. Liu, X., Cai, T. Research on the forecast model of coal mine roof based on image processing. Paper Presented at 2008 International Conference on Computer Science and Software Engineering, Wuhan, China, 12-14 December, 2008. Long, J., Shelhamer, E., Darrell, T., et al. Fully convolutional networks for semantic segmentation. Paper Presented at 2015 IEEE Conference on Computer Vision and Pattern Recognition...

A wide variety of vehicle detection approaches using deep convolutional neural network (CNN) have achieved great success in recent years. However, existing CNN-based feature extraction algorithms, especially residual network, cannot obtain powerful semantic information in vehicle detection task, and thus suffer from the problem of missing detection, error detection or repeated detection. In this paper, we present a connect-and-merge convolutional neural network (CMNet) for fast detecting vehicles in complex scenes. First, we propose a connect-and-merge residual network (CMRN) for performing feature extraction. Specifically, CMRN assembles residual branches in parallel through a connect-and-merge mapping: Connect the input to the outputs of two residual branches separately (Connect), and merge the outputs of the connection as the input of the subsequent residual block (Merge), respectively. Second, we present a multi-scale prediction network (MSPN) to accurately regress the vehicle shape and classify vehicle fine-grained categories. In addition, feature maps from CMRN are merged with their corresponding upsampled features from MSPN using concatenation. Thus, we can improve the information flow of the framework and make it easy to train. Experimental results on the KITTI dataset and the UA-DETRAC dataset demonstrate that CMNet can achieve efficient detection performance in terms of accuracy and speed for real-world traffic surveillance data.

In recent years, significant progress has been witnessed in the field of deep learning-based object detection. As a subtask in the field of object detection, traffic sign detection has great potential for development. However, the existing object detection methods for traffic sign detection in real-world scenes are plagued by issues such as the omission of small objects and low detection accuracies. To address these issues, a traffic sign detection model named YOLOv7-Traffic Sign (YOLOv7-TS) is proposed based on sub-pixel convolution and feature fusion. Firstly, the up-sampling capability of the sub-pixel convolution integrating channel dimension is harnessed and a Feature Map Extraction Module (FMEM) is devised to mitigate the channel information loss. Furthermore, a Multi-feature Interactive Fusion Network (MIFNet) is constructed to facilitate enhanced information interaction among all feature layers, improving the feature fusion effectiveness and strengthening the perception ability of small objects. Moreover, a Deep Feature Enhancement Module (DFEM) is established to accelerate the pooling process while enriching the highest-layer feature. YOLOv7-TS is evaluated on two traffic sign datasets, namely CCTSDB2021 and TT100K. Compared with YOLOv7, YOLOv7-TS, with a smaller number of parameters, achieves a significant enhancement of 3.63% and 2.68% in the mean Average Precision (mAP) for each respective dataset, proving the effectiveness of the proposed model.

Currently, many real-time semantic segmentation networks aim for heightened accuracy, inevitably leading to increased computational complexity and reduced inference speed. Therefore, striking a balance between accuracy and speed has emerged as a crucial concern in this domain. To address these challenges, this study proposes a dual-branch fusion network with multiscale atrous pyramid pooling aggregate contextual features for real-time semantic segmentation (MAFNet). The first key component, the semantics guide spatial-details module (SGSDM) not only facilitates precise boundary extraction and fine-grained classification, but also provides semantic-based feature representation, thereby enhancing support for spatial analysis and decision boundaries. The second component, the multiscale atrous pyramid pooling module (MSAPPM), is designed by combining dilation convolution with feature pyramid pooling operations at various dilation rates. This design not only expands the receptive field, but also aggregates rich contextual information more effectively. To further improve the fusion of feature information generated by the dual-branch, a bilateral fusion module (BFM) is introduced. This module employs cross-fusion by calculating weights generated by the dual-branch to balance the weight relationship between the dual branches, thereby achieving effective feature information fusion. To validate the effectiveness of the proposed network, experiments are conducted on a single A100 GPU. MAFNet achieves a mean intersection over union (mIoU) of 77.4% at 70.9 FPS on the Cityscapes test dataset and 77.6% mIoU at 192.5 FPS on the CamVid test dataset. The experimental results conclusively demonstrated that MAFNet effectively strikes a balance between accuracy and speed.