基于TrueSkill排序与深度学习的绿色工地主观视觉感知预测

doi:10.11996/JG.j.2095-302X.2026030641

图学学报 ›› 2026, Vol. 47 ›› Issue (3): 641-652.DOI: 10.11996/JG.j.2095-302X.2026030641

基于TrueSkill排序与深度学习的绿色工地主观视觉感知预测

卢德辉¹, 宋琢¹, 黄志超¹, 田时雨¹, 李慧敏², 田茂³, 邓逸川²^,⁴()

¹ 广州一建建设集团有限公司，广东广州 510060
² 华南理工大学土木与交通学院，广东广州 510641
³ 不列颠哥伦比亚大学土木工程系，不列颠哥伦比亚温哥华 V6T 1Z4
⁴ 亚热带建筑科学国家重点实验室，广东广州 510641

收稿日期:2025-11-12 接受日期:2026-02-09 出版日期:2026-06-30 发布日期:2026-06-30
通讯作者:邓逸川，E-mail：ctycdeng@scut.edu.cn
基金资助:
国家自然科学基金(52308314);广东省自然科学基金-青年提升项目(2023A1515030169);广东省住房和城乡建设厅科技创新计划(20250305J0004);广州市建筑集团有限公司科技计划项目([2023]-KJ008);广州市科技计划项目(2025A04J5216)

Subjective visual perception prediction of green construction sites based on TrueSkill ranking and deep learning

LU Dehui¹, SONG Zhuo¹, HUANG Zhichao¹, TIAN Shiyu¹, LI Huimin², TIAN Mao³, DENG Yichuan²^,⁴()

¹ Guangzhou First Construction Group Co. Ltd., Guangzhou Guangdong 510060, China
² School of Civil Engineering and Transportation, South China University of Technology, Guangzhou Guangdong 510641, China
³ Department of Civil Engineering, University of British Columbia, Vancouver, British Columbia V6T 1Z4, Canada
⁴ State Key Laboratory of Subtropical Architecture Science, Guangzhou Guangdong 510641, China

Received:2025-11-12 Accepted:2026-02-09 Published:2026-06-30 Online:2026-06-30
Contact: DENG Yichuan，E-mail：ctycdeng@scut.edu.cn
Supported by:
National Natural Science Foundation of China(52308314);Youth Enhance Project of Natural Science Foundation of Guangdong Province(2023A1515030169);Technology Innovation Program of Guangdong Provincial Department of Housing and Urban-Rural Development(20250305J0004);Technology Program Project of Guangzhou Municipal Construction Group CO. LTD.([2023]-KJ008);Guangzhou Science and Technology Program Project(2025A04J5216)

摘要/Abstract

摘要：

可持续发展是应对全球环境与能源挑战的核心路径，建筑施工行业已逐步推行绿色施工实践，但施工环境特征与人类主观感知之间的内在关联尚未得到系统探究，导致绿色施工在安全性、美观性与效率优化方面缺乏针对性理论支撑。为明确人类对施工环境的感知规律以指导绿色施工实践优化，通过在线众包平台采集受试者关于施工场景图像的随机匹配组合的偏好数据，采用微软TrueSkill系统对感知偏好结果进行量化排序。针对施工场景中的关键视觉元素，运用感兴趣区域(ROI)分析法开展特征提取与解析，同时选取并训练卷积神经网络(CNN)模型以实现感知质量的自动化预测。结果显示，施工现场在上述感知维度的主观感知与特定视觉特征存在显著统计学关联：现场建材堆放整齐度、地面整洁度与设备使用情况分别影响安全性、美观性与效率性感知。由于前2类视觉特征同属施工场地有序度的外在体现，故而安全性和美观性感知存在强正相关性，而效率性感知和另2类感知间均未显示相关性。研究创新性地提出众包技术与深度学习相结合的施工环境评估框架，验证了不同群体对施工场景视觉感知的一致性特征，明确了文明施工现场需满足的特定视觉标准，为施工环境感知质量的自动化评价体系构建提供了基准依据。

关键词: 绿色施工, 感知评价, 深度学习

Abstract:

Sustainable development serves as a core approach to addressing global environmental and energy challenges, while the construction industry has gradually implemented green construction practices, but the inherent connection between construction environment characteristics and human subjective perception has not yet been systematically explored. This gap results in a lack of targeted theoretical support for the optimization of green construction in terms of safety, aesthetics, and efficiency. To clarify the laws governing human perception of construction environments and thereby guide the optimization of green construction practices, this research collected preference data from participants regarding randomly matched combinations of two typical construction scenario images through an online crowdsourcing platform. The Microsoft TrueSkill system was employed to quantitatively rank the perceptual preference results. For the key visual elements in the construction scenarios, the Region of Interest (ROI)-based analysis method was used for feature extraction and interpretation. Meanwhile, several Convolutional Neural Networks (CNN) were selected and trained to enable the automated prediction of perceptual quality. The results indicate that there is a significant statistical correlation between the subjective perception of construction sites in the perceptual dimensions mentioned above and specific visual features: the stacking neatness of on-site building materials, ground cleanliness and equipment usage affect the perception about safety, aesthetics and efficiency, respectively. Due to the fact that the first two types of visual features are both external manifestations of the construction site orderiness, the perception of safety and aesthetics exhibits a strong positive correlation, while no correlation was observed between efficiency perception and the other two types of perception. An innovative construction environment evaluation framework was proposed, integrating crowdsourcing technology with deep learning. The consistency of visual perception of construction scenarios across different groups was clarified, and the specific visual standards required for civilized construction sites were clarified, providing a benchmark foundation for the establishment of an automated evaluation system for the perceptual quality of construction environments.

Key words: green construction, perceptual evaluation, deep learning

中图分类号:

卢德辉, 宋琢, 黄志超, 田时雨, 李慧敏, 田茂, 邓逸川. 基于TrueSkill排序与深度学习的绿色工地主观视觉感知预测[J]. 图学学报, 2026, 47(3): 641-652.

LU Dehui, SONG Zhuo, HUANG Zhichao, TIAN Shiyu, LI Huimin, TIAN Mao, DENG Yichuan. Subjective visual perception prediction of green construction sites based on TrueSkill ranking and deep learning[J]. Journal of Graphics, 2026, 47(3): 641-652.

图/表 12

图1 技术路线示意图

Fig. 1 Schematic diagram of technical route

表1 感知指标列表

Table 1 Perception indicator list

感知指标	目的
安全 (Safety)	测量观察施工现场时的安全感程度
美观 (Beauty)	测量施工现场的美观程度
效率 (Efficiency)	测量现场工人合作效率的感受程度

图2 在线问卷网站

Fig. 2 Online questionnaire website

图3 TrueSkill算法中的高斯分布模型

Fig. 3 Gaussian distribution model in TrueSkill algorithm

图4 卷积神经网络架构

Fig. 4 Convolutional neural network architecture

图5 TrueSkill收敛过程

Fig. 5 TrueSkill convergence process

图6 TrueSkill排序结果示例图像((a) 安全性；(b) 美观性；(c) 效率性)

Fig. 6 Example image of TrueSkill ranking results ((a) Safety; (b) Aesthetics; (c) Efficiency)

图7 基于人口统计特征的Q-Score相关性检验结果((a) 男性与女性；(b) 中位数年龄上下；(c) 建筑行业与其他行业)

Fig. 7 Q-Score correlation test results based on demographic characteristics ((a) Men and women; (b) Around the median age; (c) Construction industry vs. other industries)

图8 ROI数据示例((a) 安全性；(b) 美观性；(c) 效率性)

Fig. 8 ROI data example ((a) Safety; (b) Aesthetics; (c) Efficiency)

表2 基于CNN的感知指标预测准确率

Table 2 CNN-based prediction accuracy on perception indicators

感知维度	AlexNet	VGG16	GoogleNet	LeNet
安全性	0.65	0.68	0.66	0.63
美观性	0.63	0.70	0.66	0.62
效率性	0.69	0.63	0.64	0.60

表3 感知质量交叉预测结果

Table 3 Cross-prediction results in perception quality

CNN架构	安全性	美观性	效率性
VGG16 (安全性子集训练)	0.68	0.62	0.49
VGG16 (美观性子集训练)	0.63	0.70	0.51
AlexNet (效率子集训练)	0.52	0.51	0.69

图9 跨感知维度Q-Score散点图分析((a) 安全性与美观性；(b) 安全性与效率性；(c) 美观性与效率性)

Fig. 9 Cross-perceptual dimension Q-Score scatter plot analysis ((a) Safety and aesthetics; (b) Safety and efficiency; (c) Aesthetics and efficiency)

参考文献 55

[1]	MIR M, NASIRZADEH F, ZAKERI M, et al. Assessing neural markers of attention during exposure to construction noise using machine learning classification of electroencephalogram data[J]. Building and Environment, 2024, 261: 111754. DOI URL
[2]	ZHANG F, LU Y, WANG Y P, et al. Study on dust pollution law and chemical dust suppression technology of non hard pavement in urban construction sites[J]. Building and Environment, 2023, 229: 109938. DOI URL
[3]	ZHONG B T, GUO J D, ZHANG L, et al. A blockchain-based framework for on-site construction environmental monitoring: proof of concept[J]. Building and Environment, 2022, 217: 109064. DOI URL
[4]	OBIOZO R N, SMALLWOOD J J. Biophilic construction site model: enhancing the motivational and humanistic value of the green construction site[J]. Journal of Construction Engineering and Management, 2015, 141(3): 05014018. DOI URL
[5]	FU H L, XIA Z J, TAN Y B, et al. Influence of cues on the safety hazard recognition of construction workers during safety training: evidence from an eye-tracking experiment[J]. Journal of Civil Engineering Education, 2024, 150(1): 04023009. DOI URL
[6]	NEWTON S. Measuring the perceptual, physiological and environmental factors that impact stress in the construction industry[J]. Construction Innovation: Information Process Management, 2024, 24(3): 684-701.
[7]	JEBELLI H, HWANG S, LEE S. EEG-based workers’ stress recognition at construction sites[J]. Automation in Construction, 2018, 93: 315-324. DOI URL
[8]	YANG L, HONGYU C. Research on factors influencing total carbon emissions of construction based on structural equation modeling: a case study from China[J]. Building and Environment, 2025, 275: 112396. DOI URL
[9]	XU H J, KIM J I, CHEN J Y. Improved framework for estimating carbon emissions from prefabricated buildings during the construction stage: life cycle assessment and case study[J]. Building and Environment, 2025, 272: 112599. DOI URL
[10]	LI X J, WU J J, LIN C X. Decarbonizing provincial construction industry under the “dual carbon” goals: assessing reduction capacities and charting optimal pathways[J]. Building and Environment, 2025, 272: 112639. DOI URL
[11]	VAN TAM N, QUYNH T T H, TOAN N Q. How Vietnam can achieve net-zero carbon emissions in construction and built environment by 2050: an integrated AHP and DEMATEL approach[J]. Building and Environment, 2025, 274: 112752. DOI URL
[12]	GU X R, CHEN C, FANG Y, et al. CECA: an intelligent large-language-model-enabled method for accounting embodied carbon in buildings[J]. Building and Environment, 2025, 272: 112694. DOI URL
[13]	DONG Y H, NG S T. A life cycle assessment model for evaluating the environmental impacts of building construction in Hong Kong[J]. Building and Environment, 2015, 89: 183-191. DOI URL
[14]	FANG W L, WU D R, LOVE P E D, et al. Physiological computing for occupational health and safety in construction: review, challenges and implications for future research[J]. Advanced Engineering Informatics, 2022, 54: 101729. DOI URL
[15]	ABAS N H, AHMAD JALANI A F, MOHD AFFANDI H. Construction stakeholders’ perceptions of occupational safety and health risks in Malaysia[J]. International Journal of Sustainable Construction Engineering and Technology, 2022, 11(1): 300-311.
[16]	GUO P, TIAN W, LI H M, et al. Global characteristics and trends of research on construction dust: based on bibliometric and visualized analysis[J]. Environmental Science and Pollution Research, 2020, 27(30): 37773-37789. DOI
[17]	ZOU Z B, ERGAN S. Impact of construction projects on urban quality of life: a data analysis method[C]//Construction Research Congress 2018. New Orleans: American Society of Civil Engineers, 2018: 34-44.
[18]	CARDOSO TEIXEIRA J M. Construction site environmental impact in civil engineering education[J]. European Journal of Engineering Education, 2005, 30(1): 51-58. DOI URL
[19]	PEDERSEN M. The tyranny of scarcity: learning and economy at the construction site[J]. Journal of Education and Work, 2014, 27(4): 392-408. DOI URL
[20]	MALLAWAARACHCHI H, SENARATNE S. Importance of quality for construction project success[EB/OL]. [2025-07-12]. http://www.civil.mrt.ac.lk/web/conference/ICSECM_2015/volume_4/Extract/SECM-15-129.pdf.
[21]	CHAN A P C, SCOTT D, LAM E W M. Framework of success criteria for design/build projects[J]. Journal of Management in Engineering, 2002, 18(3): 120-128. DOI URL
[22]	YIN Y, ZHANG Y. Environmental pollution evaluation of urban rail transit construction based on entropy weight method[J]. Nature Environment and Pollution Technology, 2021, 20(2): 819-824.
[23]	PRASCEVIC N, PRASCEVIC Z. Application of fuzzy AHP for ranking and selection of alternatives in construction project management[J]. Journal of Civil Engineering and Management, 2017, 23(8): 1123-1135. DOI URL
[24]	OLAWUMI T O, AKINRATA E B, ARIJELOYE B T. Value management - creating functional value for construction project: an exploratory study[J]. World Scientific News, 2016, 54: 40-59.
[25]	GUNARATHNA U. A review on risk management in sustainable construction[EB/OL]. [2025-07-12]. https://192.248.104.6/bitstream/handle/345/1292/ENG-039.pdf?sequence=1.
[26]	MASINGBOON M K. Relationships among communication characteristics, job performance and job satisfaction as perceived by professional nurses in government hospitals, Bangkok Metropolis[D]. Bangkok: Chulalongkorn University, 1992.
[27]	FUNG I W H, TAM V W Y, SING C P, et al. Psychological climate in occupational safety and health: the safety awareness of construction workers in south China[J]. International Journal of Construction Management, 2016, 16(4): 315-325. DOI URL
[28]	IDREES M D, HAFEEZ M, KIM J Y. Workers’ age and the impact of psychological factors on the perception of safety at construction sites[J]. Sustainability, 2017, 9(5): 745. DOI URL
[29]	LOUW L A, SCHAAP P. Categories of human risk factors which impact on the psychological fitness of construction workers: a review of the evidence[J]. Journal of Psychology in Africa, 2013, 23(4): 589-599. DOI URL
[30]	GUO H L, YU Y T, XIANG T, et al. The availability of wearable-device-based physical data for the measurement of construction workers’psychological status on site: from the perspective of safety management[J]. Automation in Construction, 2017, 82: 207-217. DOI URL
[31]	LOVE P E D, EDWARDS D J. Taking the pulse of UK construction project managers’ health: influence of job demands, job control and social support on psychological wellbeing[J]. Engineering, Construction and Architectural Management, 2005, 12(1): 88-101. DOI URL
[32]	MOHAMED S. Safety climate in construction site environments[J]. Journal of Construction Engineering and Management, 2002, 128(5): 375-384. DOI URL
[33]	OBIOZO R, SMALLWOOD J. Healing gardens for the construction site: an innovative organisational management strategy[C]// The 8th CIDB Postgraduate Conference. Johannesburg, South Africa: University of the Witwatersrand, 2014: 325.
[34]	JOKKAW N, SUTEECHARUWAT P, WEERAWETWAT P. Measurement of construction workers’ feeling by virtual environment (VE) technology for guardrail design in high-rise building construction projects[J]. Engineering Journal, 2017, 21(5): 161-177.
[35]	QIU W S, LI W J, LIU X, et al. Subjectively measured streetscape perceptions to inform urban design strategies for Shanghai[J]. ISPRS International Journal of Geo-Information, 2021, 10(8): 493. DOI URL
[36]	EWING R, HANDY S. Measuring the unmeasurable: urban design qualities related to walkability[J]. Journal of Urban Design, 2009, 14(1): 65-84. DOI URL
[37]	PARK C S, SRINIVASAN V. A survey-based method for measuring and understanding brand equity and its extendibility[J]. Journal of Marketing Research, 1994, 31(2): 271-288. DOI URL
[38]	LIN L, MOUDON A V. Objective versus subjective measures of the built environment, which are most effective in capturing associations with walking?[J]. Health & Place, 2010, 16(2): 339-348.
[39]	APPLETON J. The experience of landscape[M]. United Kingdom: Wiley, 1996: 69-74.
[40]	GRIEW P, HILLSDON M, FOSTER C, et al. Developing and testing a street audit tool using Google street view to measure environmental supportiveness for physical activity[J]. International Journal of Behavioral Nutrition and Physical Activity, 2013, 10(1): 103. DOI
[41]	SALESSES P, SCHECHTNER K, HIDALGO C A. The collaborative image of the city: mapping the inequality of urban perception[J]. PLOS One, 2013, 8(7): e68400. DOI URL
[42]	RUNDLE A G, BADER M D M, RICHARDS C A, et al. Using Google street view to audit neighborhood environments[J]. American Journal of Preventive Medicine, 2011, 40(1): 94-100. DOI PMID
[43]	SERESINHE C I, PREIS T, MOAT H S. Using deep learning to quantify the beauty of outdoor places[J]. Royal Society Open Science, 2017, 4(7): 170170. DOI URL
[44]	NAIK N, PHILIPOOM J, RASKAR R, et al. Streetscore - predicting the perceived safety of one million streetscapes[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops. New York: IEEE Press, 2014: 793-799.
[45]	DONG Z, SHEN X, LI H Q, et al. Photo quality assessment with DCNN that understands image well[C]// The 21st International Conference on Multimedia Modeling. Cham: Springer, 2015: 524-535.
[46]	ONUBI H O, YUSOF N, HASSAN A S. Adopting green construction practices: health and safety implications[J]. Journal of Engineering, Design and Technology, 2019, 18(3): 635-652. DOI URL
[47]	KARAKHAN A A. LEED-certified projects: green or sustainable?[J]. Journal of Management in Engineering, 2016, 32(5): 02516001. DOI URL
[48]	ONUBI H O, YUSOF N, HASSAN A S. Understanding the mechanism through which adoption of green construction site practices impacts economic performance[J]. Journal of Cleaner Production, 2020, 254: 120170. DOI URL
[49]	HOX J J, BOEIJE H R. Data collection, primary vs. secondary[M]//KEMPF-LEONARD K. Encyclopedia of Social Measurement. Amsterdam: Elsevier, 2005: 593-599.
[50]	STEWART N, BROWN G D A, CHATER N. Absolute identification by relative judgment[J]. Psychological Review, 2005, 112(4): 881-911. PMID
[51]	HERBRICH R, MINKA T, GRAEPEL T. TrueSkill^TM: a Bayesian skill rating system[C]// The 20th International Conference on Neural Information Processing Systems. Cambridge: MIT Press, 2007: 569-576.
[52]	MINKA T, ZAYKOV Y, TIMS J. TrueSkill^TM ranking system[EB/OL]. (2005-11-18) [2025-02-25]. https://www.microsoft.com/en-us/research/project/trueskill-ranking-system/.
[53]	DUBEY A, NAIK N, PARIKH D, et al. Deep learning the city: quantifying urban perception at a global scale[C]// The 14th European Conference on Computer Vision. Cham: Springer, 2016: 196-212.
[54]	DUAN R, DENG H, TIAN M, et al. SODA: a large-scale open site object detection dataset for deep learning in construction[J]. Automation in Construction, 2022, 142: 104499. DOI URL
[55]	ELKAPELLI S S, DAMAHE L B. A review: region of interest based image retrieval[C]// 2016 Online International Conference on Green Engineering and Technologies. New York: IEEE Press, 2016: 1-6.

基于TrueSkill排序与深度学习的绿色工地主观视觉感知预测

Subjective visual perception prediction of green construction sites based on TrueSkill ranking and deep learning

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 12

参考文献 55

相关文章 15

编辑推荐

Metrics

本文评价

[1]	李秀梅, 周正鑫, 孙军梅. 一种定位分支辅助的多任务协同图像伪造检测模型[J]. 图学学报, 2026, 47(3): 524-533.
[2]	吴文欢, 王文舒, 王舒鳌. 融合层次化双流注意力的单目深度估计方法[J]. 图学学报, 2026, 47(3): 553-563.
[3]	闫康, 曾理, 顾晓清. 基于跨域结构化深度字典学习的图像分类方法[J]. 图学学报, 2026, 47(2): 341-350.
[4]	庞敏, 李振堂, 张元, 崔晓康, 熊风光. 基于检索与变形技术的三维模型重构[J]. 图学学报, 2026, 47(2): 368-379.
[5]	董文益, 杨伟东, 唐冰慧, 王琦, 肖宏宇. 基于深度学习的肝脏局灶性病变检测方法综述[J]. 图学学报, 2026, 47(1): 1-16.
[6]	翟永杰, 王紫萱, 张祯琪, 周迅琪, 王乾铭. 融合双重注意力与加权动态卷积的车辆损伤分类模型[J]. 图学学报, 2026, 47(1): 17-28.
[7]	潘宇轩, 金锐, 刘雨, 张琳. 基于生成模型的无监督多视点立体视觉网络[J]. 图学学报, 2026, 47(1): 29-38.
[8]	酒明远, 吴国伟, 宋旭光, 李书攀, 徐明亮. 基于不确定性引导的智能强化主动学习图像分类方法[J]. 图学学报, 2026, 47(1): 47-56.
[9]	杨彪, 王学, 官铮, 龙萍. BSD-YOLO：基于动态稀疏注意力与自适应检测头的小目标车辆检测方法[J]. 图学学报, 2026, 47(1): 99-110.
[10]	琚晨, 丁嘉欣, 王泽兴, 李广钊, 管振祥, 张常有. 面向有限元法的图神经网络形函数近似方法[J]. 图学学报, 2025, 46(6): 1161-1171.
[11]	易斌, 张立斌, 刘丹楹, 唐军, 方俊俊, 李雯琦. 基于AMTA-Net的卷制过程激光打孔通风率预测模型[J]. 图学学报, 2025, 46(6): 1224-1232.
[12]	薄文, 琚晨, 刘维青, 张焱, 胡晶晶, 程婧晗, 张常有. 基于退化感知时序建模的装备维保时机预测方法[J]. 图学学报, 2025, 46(6): 1233-1246.
[13]	赵振兵, 欧阳文斌, 冯烁, 李浩鹏, 马隽. 基于类内稀疏先验与改进YOLOv8的绝缘子红外图像检测方法[J]. 图学学报, 2025, 46(6): 1247-1256.
[14]	贺蒙蒙, 张小艳, 李洪安. 基于Mamba结构的轻量级皮肤病变图像分割网络[J]. 图学学报, 2025, 46(6): 1257-1266.
[15]	李星辰, 李宗民, 杨超智. 基于可信伪标签微调的测试时适应算法[J]. 图学学报, 2025, 46(6): 1292-1303.