集装箱堆场作为交接保管集装箱的重要场所，其内存放的集装箱数量信息对于堆场的管理以及数据统计工作十分重要。铁路场站堆场相比其它物流堆场而言，周转速度更快，更容易发生作业司机因疏忽而滞后点击作业指令的情况，从而导致集装箱账面数量与堆场内实际数量之间存在误差。等后期发现误差存在时，核查的工作量巨大。为了降低铁路场站集装箱堆场的管理成本，及时发现账面出错的情况，亟需研发一款成本较低、检测性能较好的集装箱数量识别系统。可以实时对堆场内的实际集装箱数量进行反馈，降低后期核查成本，提高铁路场站的工作效率。

    为了低成本、高效地解决铁路场站内集装箱数量自动核查问题，本文使用堆场监控设备拍摄的图像（以下简称“监控视图”）实现了一种铁路场站堆场的集装箱数量计算方法，主要研究内容如下：

    （1）在铁路场站堆场内的监控图像中，实现对集装箱目标的实时检测。首先针对监控视图下的露天铁路场站图像可能存在的包括低照度和局部过曝光问题进行了分析研究，并提出了解决方案。然后对目标检测算法展开了研究，选择采用在速度和精度上都较高的YOLOv7（You Only Look Once）目标检测算法，为了更好的适用于识别集装箱，选择对损失函数进行了替换；在此基础上，针对目标检测过程中可能出现的包括错检以及预测框重叠在内的问题进行了处理。最后设计了实验并分析结果，验证了集装箱检测算法的有效性，为后续集装箱堆存推算算法的设计提供了良好的基础。

    （2）构建铁路场站集装箱堆存模型，设计了一种基于监控视图的集装箱堆存自动推算算法（Calculation Algorithm of Container Stockpile Based on Image, CACSBI）。针对集装箱检测算法无法识别到受遮挡集装箱的问题，设计了一种集装箱堆存自动推算算法CACSBI，实现了对铁路场站堆场内集装箱的自动计数。首先根据集装箱在堆场中的堆存形式构建了集装箱堆存模型，并以此模型为基础，结合集装箱预测框信息和堆存规则，计算集装箱所在位置，进而确定了可见集装箱的堆存情况；然后对监控视图下可能存在的遮挡情况进行了分析，并设计了根据可见集装箱推算受遮挡集装箱的方案，进而有效地推导出了堆场内集装箱的实际堆存；最终根据推算流程设计实验，验证了集装箱自动推算算法的有效性和可行性。

    （3）设计并实现了一个集装箱数量自动计算系统。为了满足低成本需求，选择采用铁路场站堆场内现有的监控设备，经实验测试，每个监控设备对4排20贝范围内的集装箱检测效果较好。因为随着贝号增大，与监控设备的距离越远，预测结果的置信度就越低。基于此，设计了集装箱计数界面，将预测的实际堆存与集装箱管理系统记录的账面堆存进行对比，将不一致集装箱的位置及当前位置的预测置信度显示在界面上，根据置信度的数值给出相应的核查建议。最后在计数系统界面上设置了集装箱管理系统的跳转按钮，当发现账面堆存信息有误时，可以方便地跳转至管理系统进行修改。该计数系统可以帮助铁路场站的工作人员更好地管理集装箱信息，降低出错成本，在一定程度上提高了铁路场站堆场的作业效率。

    综上所述，本文首先对集装箱计数任务进行了深入分析，然后结合实际应用场景的特性，对目标检测算法进行了调整，并在此基础上设计了集装箱堆存推算算法。在自制数据集上的实验表明，该计数系统能够有效地完成监控视图下铁路场站堆场的集装箱计数任务，具有一定的应用价值。

[1]江泽涛, 翟丰硕, 钱艺, 等. 结合特征增强和多尺度感受野的低照度目标检测[J/OL]. 计算机研究与发展, 2022: 1-13. DOI:10.7544/issn1000-1239.20211092.

[2]李幼蛟, 卓力, 张菁, 等. 行人再识别技术综述[J]. 自动化学报, 2018, 44(09): 1554-1568.

[3]张慧, 王坤峰, 王飞跃. 深度学习在目标视觉检测中的应用进展与展望[J]. 自动化学报, 2017, 43(08): 1289-1305.

[4]Huang Y S, Jiang Z T, Lan R S, et al. Infrared Image Super-Resolution via Transfer Learning and PSRGAN[J]. IEEE Signal Processing Letters, 2021, 28: 982-986.

[5]梁杰. 安防监控系统中视频目标检测算法研究[D]. 电子科技大学, 2022.

[6]陈科圻, 朱志亮, 邓小明, 等. 多尺度目标检测的深度学习研究综述[J]. 软件学报, 2021, 32(4): 1201-1227.

[7]Deng J, Dong W, Socher R, et al. ImageNet: A large-scale hierarchical image database[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2009: 248-255.

[8]Krizhevsky A, Sutskever I, Hinton G. ImageNet Classification with Deep Convolutional Neural Networks[J]. Advances in neural Information Processing Systems, 2012, 25(2): 1097-1105.

[9]田卓钰, 马苗, 杨楷芳. 基于级联注意力与点监督机制的考场目标检测模型[J]. 软件学报, 2022, 33(7): 2633-2645.

[10]平昱恺, 黄鸿云, 江贺, 等. 目标检测模型的决策依据与可信度分析[J]. 软件学报, 2022, 33(9): 3391-3406.

[11]Girshick R, Donahue J, Darrell T, et al. Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation[J]. IEEE Computer Society, 2014: 580-587.

[12]Girshick R. Fast R-CNN[J]. Computer Science, 2015: 1440-1448.

[13]Ren S, He K, Girshick R, et al. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6): 1137-1149.

[14]Redmon J, Divvala S, Girshick R, et al. You Only Look Once: Unified, Real-Time Object Detection[J]. IEEE Computer Society, 2016: 779-788.

[15]Liu W, Anguelov D, Erhan D, et al. SSD: Single shot multibox detector[P]. In Proceedings of the European Conference on Computer Vision, 2016: 21-37.

[16]Redmon J, Farhadi A. YOLO9000: Better, faster, stronger[C]. In proceedings of the IEEE conference on Computer Vision and Pattern Recognition, 2017: 7263-7271.

[17]Redmon J, Farhadi A. Yolov3: An incremental improvement[J]. arXiv preprint, arXiv: 1804.02767, 2018: 1-6.

[18]Bochkovskiy A, Wang C Y, Liao H Y M. Yolov4: Optimal speed and accuracy of object detection[J]. arXiv preprint, arXiv:2004.10934, 2020: 1-17.

[19]Ge Z, Liu S T, Wang F, et al. YOLOX: Exceeding YOLO series in 2021[J]. arXiv preprint arXiv:2107.08430,2021: 1-7.

[20]Li C Y, Li L L, Jiang H L, et al. YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications[J]. arXiv preprint, arXiv:2209.02976, 2022: 1-7.

[21]Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[J]. arXiv preprint, arXiv:2207.02696, 2022: 1-15.

[22]Wang D, Hwang M, Jiang W C, et al. A deep learning method for counting white blood cells in bone marrow images[J]. BMC Bioinformatics, 2021, 22(5): 94-107.

[23]Jiang K L, Xie T Y, Yan R, et al. An Attention Mechanism-Improved YOLOv7 Object Detection Algorithm for Hemp Duck Count Estimation[J]. Agriculture, 2022, 12(10): 1659-1659.

[24]Wang P P, Meng F R, Donaldson P, et al. High throughput measurement of plant fitness traits with an object detection method using Faster R-CNN[J]. The New phytologist, 2022, 234(4): 1521-1533.

[25]Li Y, Zhang X, Chen D. CSRNet: Dilated Convolutional Neural Networks for Understanding the Highly Congested Scenes[C]. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 1091-1100.

[26]张宇倩, 李国辉, 雷军, 等. FF-CAM:基于通道注意机制前后端融合的人群计数[J]. 计算机学报, 2021, 44(02): 304-317.

[27]Sindagi V A, Patel V M. A Survey of Recent Advances in CNN-based Single Image Crowd Counting and Density Estimation[J]. Pattern Recognition Letters, 2018, 107: 3-16.

[28]Stahl T, Pintea S L, Gemert J C V. Divide and Count: Generic Object Counting by Image Divisions[J]. In IEEE Transactions on Image Processing, 2019, 28: 1035-1044.

[29]Tuzel O, Porikli F, Meer P. Pedestrian Detection via Classification on Riemannian Manifolds. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008, 30(10): 1713-1727.

[30]Chan A B, Vasconcelos N. Bayesian Poisson regression for crowd counting[C]. In Proceedings of IEEE International Conference on Computer Vision (ICCV). IEEE, 2009: 545-551.

[31]Chen K, Loy C C, Gong S, et al. Feature Mining for Localised Crowd Counting[C]. In Proceedings of British Machine Vision Conference (BMVC), 2012: 1-11.

[32]Lempitsky V, Zisserman A. Learning to count objects in images[C]. In Proceedings of Advances in Neural Information Processing Systems (NIPS), 2010: 1324-1332.

[33]Wang Y, Zou Y. Fast visual object counting via example-based density estimation[C]. In Proceedings of the IEEE International Conference on Image Processing (ICIP). IEEE, 2016: 3653-3657.

[34]Xu B, Qiu G. Crowd density estimation based on rich features and random projection forest[C]. In Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE, 2016: 1-8.

[35]Zhang C, Li H, Wang X, et al. Cross-scene crowd counting via deep convolutional neural networks[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015: 833-841.

[36]Wang C, Zhang H, Yang L, et al. Deep people counting in extremely dense crowds[C]. In Proceedings of the ACM International Conference on Multimedia (ACMMM). ACM, 2015: 1299-1302.

[37]Fu M, Xu P, Li X, et al. Fast crowd density estimation with convolutional neural networks[J]. Engineering Applications of Artificial Intelligence, 2015, 43: 81-88.

[38]Boominathan L, Kruthiventi S S, Babu R V. Crowdnet: A deep convolutional net-work for dense crowd counting[C]. In Proceedings of the ACM International Conference on Multimedia (ACMMM), 2016: 640-644.

[39]Sam D B, Surya S, Babu R V. Switching convolutional neural network for crowd counting[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017: 4031-4039.

[40]彭贤, 彭玉旭, 汤强, 等. 基于单列多尺度卷积神经网络的人群计数[J]. 计算机科学, 2020, 47(04): 150-156.

[41]Jiang X, Xiao Z, Zhang B, et al. Crowd Counting and Density Estimation by Trellis Encoder-Decoder Networks[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 6133-6142.

[42]Xu C, Qiu K, Fu J, et al. Learn to Scale: Generating Multipolar Normalized Density Maps for Crowd Counting[C]. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 8382-8390.

[43]Sindagi V A, Patel V M. Multi-Level Bottom-Top and Top-Bottom Feature Fusion for Crowd Counting[C]. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 1002-1012.

[44]Sindagi V A, Yasarla R, Patel V M. Pushing the Frontiers of Unconstrained Crowd Counting: New Dataset and Benchmark Method[C]. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 1221-1231.

[45]Shen Z, Xu Y, Ni B, et al. Crowd counting via adversarial cross-scale consistency pursuit[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018: 5245-5254.

[46]Xiong H, Lu H, Liu C, et al. From Open Set to Closed Set: Counting Objects by Spatial Divide-and-Conquer[C]. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 8362-8371.

[47]Liu L, Qiu Z, Li G, et al. Crowd Counting with Deep Structured Scale Integration Network[C]. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 1774-1783.

[48]Ahmadi M, Xu Z, Wang X, et al. Fast Multi Object Detection and Counting by YOLO V3[C]. 2021 China Automation Congress (CAC), 2021: 7401-7404.

[49]Laradji I H, Rostamzadeh N, Pinheiro P O, et al. Where are the blobs: Counting by localization with point supervision[C]. In Proceedings of the European Conference on Computer Vision (ECCV), 2018: 547-562.

[50]Liu W, Salzmann M, Fua P. Context-Aware Crowd Counting[C]. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019: 5099-5108.

[51]Zhu A, Zheng Z, Huang Y, et al. CACrowdGAN: Cascaded Attentional Generative Adversarial Network for Crowd Counting[J]. IEEE Transactions on Intelligent Transportation Systems, 2021, 99: 1-13.

[52]Wang Y, Liu M, Yi H, et al. High-density Image Object Counting Network Based on Spatial Context and Channel Attention: Take crowds and vehicles in a statistically dense scene as an example[C]. In Proceedings of the International Conference on Electronic Information Technology and Computer Engineering, 2022: 1592-1596.

[53]李姚舜, 刘黎志. 嵌入注意力机制的轻量级钢筋检测网络[J]. 计算机应用, 2022, 42(09): 2900-2908.

[54]Cai Y Q, Du D, Zhang L, et al. 2020. Guided Attention Network for Object Detection and Counting on Drones[C]. In Proceedings of the ACM International Conference on Multimedia (MM 20), Association for Computing Machinery, 2020: 709-717.

[55]Zhang Y, Shu B, Xue Z, et al. Prediction and counting of field wheat based on LC-DcVgg[C]. In Proceedings of the International Conference on Mathematics and Artificial Intelligence (ICMAI 2020), 2020: 209-213.

[56]Laradji I H, Pardinas R, Rodriguez P, et al. Looc: Localize Overlapping Objects with Count Supervision[C]. 2020 IEEE International Conference on Image Processing (ICIP), 2020: 2316-2320.

[57]Jiang N, Yu F. A refinement on detection in cell counting[C]. 2021 IEEE International Conference on Consumer Electronics and Computer Engineering (ICCECE), 2021: 306-309.

[58]Tongmonwit B, Tamee K. Increasing Efficiency in Counting Number of Motorcycle with Object Relation Matching[J]. 2022 Joint International Conference on Digital Arts, Media and Technology with ECTI Northern Section Conference on Electrical, Electronics, Computer and Telecommunications Engineering, 2022: 314-318.

[59]Zhang J, Zhang S, Wang S, et al. Research on Target Counting Based on Improved YOLOv5 and SORT Algorithms[C]. In Proceedings of the International Conference on Advances in Image Processing (ICAIP 22), 2023: 109-113.

[60]薛卓群, 席志红. 基于修正饱和度特征的过曝光区域检测算法[J]. 电子技术与软件工程, 2019, 157(11): 68-70.

[61]Guo D, Cheng Y, S. Zhuo, T. Sim. Correcting over-exposure in photographs[C]. 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2010: 515-521.

[62]Fattal R, Lischinski D, Werman M. Gradient domain high dynamic range Compression[C]. Conference on Computer Graphics & Interactive Techniques. ACM, 2002, 21(3): 249-256.

[63]Land E H, Mccann J J. Lightness and retinex theory[J]. Journal of the Optical Society of America, 1971, 61(1): 1-11.

[64]赵征鹏, 李俊钢, 普园媛. 基于卷积神经网络的Retinex低照度图像增强[J]. 计算机科学, 2022, 49(06): 199-209.

[65]Zhang K, Zuo W, Chen Y, et al. Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising[J]. IEEE Transactions on Image Processing, 2016, 26(7): 3142-3155.

[66]Brauwers G, Frasincar F. A general survey on attention mechanisms in deep learning[J]. IEEE Transactions on Knowledge and Data Engineering, arXiv:2203.14263, 2021: 1-20.

[67]Gevorgyan Z. SIoU Loss: More Powerful Learning for Bounding Box Regression[J]. arXiv preprint arXiv:2205.12740, 2022: 1-12.