深度学习滑坡识别算法中样本不平衡问题的研究

doi:10.13474/j.cnki.11-2246.2024.0503

测绘通报 ›› 2024, Vol. 0 ›› Issue (5): 12-18.doi: 10.13474/j.cnki.11-2246.2024.0503

• 滑坡灾害识别 • 上一篇

深度学习滑坡识别算法中样本不平衡问题的研究

王丽霞¹, 喜文飞^1,2, 史正涛^1,2, 赵子龙³, 钱堂慧¹, 赵磊¹, 马奕捷¹

1. 云南师范大学地理学部, 云南昆明 650500;
2. 云南省高原地理过程与环境变化重点实验室, 云南昆明 650500;
3. 云南海钜地理信息技术有限公司, 云南昆明 650000

收稿日期:2023-09-01 发布日期:2024-06-12
作者简介:王丽霞(1995—),女,硕士生,主要研究方向为深度学习遥感影像滑坡灾害识别。E-mail:690125218@qq.com
基金资助:
滇中引水工程(滇中高原区)高分综合应用示范(89-Y50-G31-9001-22/23-05);云南省重大科技专项(202202AD080010);云南师范大学研究生科研创新基金(YJSJJ23-B152);云南省高校高烈度地震山区交通走廊工程地质病害早期快速判识与防空重点实验室开放基金(KLGDTC-2021-02)

Study on sample unbalance in landslide recognition algorithm based on depth learning

WANG Lixia¹, XI Wenfei^1,2, SHI Zhengtao^1,2, ZHAO Zilong³, QIAN Tanghui¹, ZHAO Lei¹, MA Yijie¹

1. Faculty of Geography, Yunnan Normal University, Kunming 650500, China;
2. Key Laboratory of Geographical Process and Environmental Change of Yunnan Plateau, Kunming 650500, China;
3. Yunnan Haiju Geographic Information Technology Co., Ltd., Kunming 650000, China

Received:2023-09-01 Published:2024-06-12

摘要/Abstract

摘要： 山体滑坡是一种常见的地质灾害,一旦发生会给自然生态系统和人类造成重大财产损失和人员伤亡,如何快速准确获取滑坡信息对防灾减灾至关重要。传统的深度学习方法对滑坡样本质量依赖性大,但现有的样本质量参差不齐,极少考虑滑坡样本不平衡问题对深度学习模型性能的影响。针对如何通过改善样本质量提升模型精度的问题,本文从样本质量出发,提出了一种基于多源不平衡样本的Faster R-CNN滑坡目标检测方法,通过对多种不平衡样本的集成训练,研究不同样本对模型综合性能的影响。结果表明:①在困难样本不平衡下模型的准确率为85.16%,F1值为0.69,精确率为56.96%,召回率为86.58%、漏检率为0.33,通过强化样本质量后准确率提升2.04%,精确率提升4.29%,召回率提升1.71%,漏检率降低0.04;②在正负样本不平衡下模型准确率为96.03%,F1值为0.78,精确率为64.50%,召回率为97.15%、漏检率为0.09,通过增加困难样本参与训练后,准确率下降8.45%,精确率下降6.93%,召回率下降7.25%,漏检率提升0.18。困难样本对模型综合性能影响更大,通过提高这部分样本质量可以提升模型检测精度。因此,本文提出的方法为解决深度学习中滑坡数据样本不平衡问题提供了参考。

关键词: 深度学习, 滑坡检测, Faster R-CNN, 不平衡样本, GF-2遥感影像

Abstract: Landslides are a common geological disaster that can cause significant property losses and casualties to natural ecosystems and humans once they occur. How to quickly and accurately obtain landslide information is crucial to disaster prevention and mitigation. Traditional deep learning methods depend heavily on the quality of landslide samples, but the quality of existing samples is uneven, and the impact of landslide sample imbalance on the performance of deep learning models is rarely considered. Aiming at the problem of how to improve model accuracy by improving sample quality, this paper proposes a Faster R-CNN landslide target detection method based on multi-source unbalanced samples starting from sample quality. By conducting integrated training on a variety of imbalanced samples, the impact of different samples on the comprehensive performance of the model is studied. The results show that:①The accuracy rate of the model is 85.16%, F1 score of 0.69, precision of 56.96%, recall of 86.58%, and the missed detection rate is 0.33 under the imbalance of difficult samples. After strengthening the sample quality, the accuracy rate increases by 2.04%. The precision increased by 4.29%, the recall rate increased by 1.71%, and the missed detection rate decreased by 0.04. ②Under the imbalance of positive and negative samples, the accuracy rate of the model is 96.03%,F1 score of 0.78, precision of 64.50%, recall of 97.15%, and the missed detection rate is 0.09. After adding difficult samples to participate in the training, the accuracy rate drops by 8.45%. The rate dropped by 6.93%, the recall rate dropped by 7.25%, and the missed detection rate increased by 0.18. Difficult samples have a greater impact on the overall performance of the model. By improving the quality of these samples, the model detection accuracy can be improved. Therefore, the method proposed in this article provides a reference for solving the problem of landslide data sample imbalance in deep learning.

Key words: deep learning, landslide detection, faster R-CNN, imbalanced samples, GF-2

中图分类号:

P237

王丽霞, 喜文飞, 史正涛, 赵子龙, 钱堂慧, 赵磊, 马奕捷. 深度学习滑坡识别算法中样本不平衡问题的研究[J]. 测绘通报, 2024, 0(5): 12-18.

WANG Lixia, XI Wenfei, SHI Zhengtao, ZHAO Zilong, QIAN Tanghui, ZHAO Lei, MA Yijie. Study on sample unbalance in landslide recognition algorithm based on depth learning[J]. Bulletin of Surveying and Mapping, 2024, 0(5): 12-18.

参考文献

[1] 杨迎冬,晏祥省,王宇,等.云南省地质灾害特征及形成规律研究[J].灾害学,2021,36(3): 131-139.
[2] 黄润秋.20世纪以来中国的大型滑坡及其发生机制[J].岩石力学与工程学报,2007,26(3): 433-454.
[3] 王宇.云南省地质灾害防治与研究历史评述[J].灾害学,2019,34(3): 134-139.
[4] 牛朝阳,高欧阳,刘伟,等.光学遥感图像滑坡检测研究进展[J].航天返回与遥感,2023,44(3): 133-144.
[5] MIELE P,DI NAPOLI M,GUERRIERO L,et al.Landslide awareness system (LAwS) to increase the resilience and safety of transport infrastructure: the case study of pan-American highway(Cuenca-Ecuador)[J].Remote Sensing,2021,13(8): 1564.
[6] 许冲,戴福初,徐锡伟.汶川地震滑坡灾害研究综述[J].地质论评,2010,56(6):860-874.
[7] 王庆雅,张锦.高分辨率遥感影像滑坡纹理特征提取方法[J].太原理工大学学报,2021,52(4): 547-556.
[8] JI Shunping,YU Dawen,SHEN Chaoyong,et al.Landslide detection from an open satellite imagery and digital elevation model dataset using attention boosted convolutional neural networks[J].Landslides,2020,17(6): 1337-1352.
[9] PRADHAN B,LEE S.Regional landslide susceptibility analysis using back-propagation neural network model at Cameron Highland,Malaysia[J].Landslides,2010,7(1): 13-30.
[10] MYLES A J,FEUDALE R N,LIU Y,et al.An introduction to decision tree modeling[J].Journal of Chemometrics,2004,18(6): 275-285.
[11] MOUNTRAKIS G,IM J,OGOLE C.Support vector machines in remote sensing: a review[J].ISPRS Journal of Photogrammetry and Remote Sensing,2011,66(3): 247-259.
[12] BREIMAN L.Random forests[J].Machine Learning,2001,45(1): 5-32.
[13] GHORBANZADEH O,BLASCHKE T,GHOLAMNIA K,et al.Evaluation of different machine learning methods and deep-learning convolutional neural networks for landslide detection[J].Remote Sensing,2019,11(2): 21.
[14] 巨袁臻,许强,金时超,等.使用深度学习方法实现黄土滑坡自动识别[J].武汉大学学报(信息科学版),2020,45(11): 1747-1755.
[15] GARCIA G P B,GROHMANN C H,SOARES L P,et al.Relict landslide detection using deep-learning architectures for image segmentation in rainforest areas: a new framework[J].International Journal of Remote Sensing,2023,44(7): 2168-2195.
[16] YU B,WANG N,XU C,et al.A network for landslide detection using large-area remote sensing images with multiple spatial resolutions[J].Remote Sensing,2022,14: 5759.
[17] CHENG Gong,GUO Lei,ZHAO Tianyun,et al.Automatic landslide detection from remote-sensing imagery using a scene classification method based on BoVW and pLSA[J].International Journal of Remote Sensing,2013,34(1): 45-59.
[18] LI B,LIU Y,WANG X.Gradient Harmonized Single-stage Detector[C]//Proceedings of the 33th AAAI Conference on Artificial Intelligence and Thirty-First Innovative Applications of Artificial Intelligence Conference and Ninth AAAI Symposium on Educational Advances in Artificial Intelligence.[S.l.]:AAAI,2019.
[19] XUERONG C,ZHAO C,XI J,et al.Deep learning method of landslide inventory map with imbalanced samples in optical remote sensing[J].Remote Sensing,2022,14: 5517.
[20] 刘莉,张丽萍,李萍,等.金沙江巧家巨型古滑坡发育特征及其形成条件[J].人民长江,2022,53(1): 118-125.
[21] LI Tiantao,PEI Xiangjun,HUANG Runqiu,et al.The formation and evolution of the Qiaojia pull-apart basin,North Xiaojiang Fault Zone,Southwest China[J].Journal of Mountain Science,2016,13(6): 1096-1106.
[22] JOHNSON J M,KHOSHGOFTAAR T M.Survey on deep learning with class imbalance[J].Journal of Big Data,2019,6(1): 27.
[23] 贺宇哲,何宁,张人,等.面向深度学习目标检测模型训练不平衡研究[J].计算机工程与应用,2022,58(5): 172-178.
[24] WU X,SAHOO D,HOI S C H.Recent advances in deep learning for object detection[J].Neurocomputing,2020,396:39-64.
[25] MA C,BAI X Z.Analysis of false alarm and missing alarm in conjunction assessment of space objects[J].Lecture Notes in Electrical Engineering,2014,323: 477-487.

深度学习滑坡识别算法中样本不平衡问题的研究

Study on sample unbalance in landslide recognition algorithm based on depth learning

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	林泊锟, 丁勇, 李登华. 一种基于深度学习与图像局部特征提取的边坡异常监测技术[J]. 测绘通报, 2024, 0(4): 23-28.
[2]	庞敏. 基于深度学习的国产卫片图斑提取[J]. 测绘通报, 2024, 0(4): 124-128,134.
[3]	陈昶, 张岩, 李坤衡, 杨润泽, 张俊彬, 梁彦荣. 基于高分七号影像自动提取东北黑土区侵蚀沟的方法[J]. 测绘通报, 2024, 0(3): 1-7.
[4]	张兰兰, 王红雷. 基于深度学习的耕地非农化遥感监测与时空分析——以开阳县为例[J]. 测绘通报, 2024, 0(3): 13-18.
[5]	王鸿雁, 车向红, 徐辛超, 徐胜华, 李洪胜. 利用DeepLabv3+模型提取分析街景图像绿视率——以北京三环内为例[J]. 测绘通报, 2024, 0(3): 88-94.
[6]	马锦山, 贾国焕, 张赛, 张炯. 基于多源高分辨率遥感影像的典型自然资源要素提取[J]. 测绘通报, 2024, 0(3): 123-126,150.
[7]	王红雷, 严文璞. 基于深度学习技术的贵州省耕地占用遥感监测[J]. 测绘通报, 2024, 0(3): 134-139.
[8]	刘丽, 李士垚, 王润, 刘少宇, 宋永飞, 牛瑞卿. 联合深度学习与面向对象分析的卫宁北山露天矿山采场信息提取[J]. 测绘通报, 2024, 0(2): 51-57.
[9]	黄应华, 董振川, 李昊, 陈壮, 刘长睿, 张献州. 城市竣工测绘典型要素语义分割PointNet++深度学习模型适用性分析[J]. 测绘通报, 2024, 0(2): 85-89.
[10]	刘亦博, 吴传文, 周宗锟, 陈华. 融合SuperGlue方法的深度单目视觉惯性导航算法[J]. 测绘通报, 2024, 0(2): 113-117.
[11]	魏德宏, 禤键豪, 杨嘉伟, 张兴福, 余旭. 基于深度BP/ELMAN神经网络的山区GNSS高程转换精度分析[J]. 测绘通报, 2023, 0(9): 113-116,143.
[12]	张宝安, 高小龙, 金仔燕, 马兰花. 甘肃省地表要素遥感解译样本库建设与应用[J]. 测绘通报, 2023, 0(9): 144-149.
[13]	胡荣明, 李鑫, 竞霞, 武建强, 魏青博. YOLOv7在探地雷达B-Scan图像解译中的应用[J]. 测绘通报, 2023, 0(8): 29-33.
[14]	周宝定, 谢沛瑶, 郭文浩, 毛庆洲. 基于激光点云灰度图像的隧道渗水病害检测[J]. 测绘通报, 2023, 0(8): 34-39,90.
[15]	张戬, 高雅. 深度学习遥感影像解译技术在耕地保护中的应用[J]. 测绘通报, 2023, 0(8): 142-145.