Transfer learning and deep convolutional neural networks for safety guardrail detection in 2D images

被引:205
作者
Kolar, Zdenek [1 ]
Chen, Hainan [1 ]
Luo, Xiaowei [1 ]
机构
[1] City Univ Hong Kong, Dept Architecture & Civil Engn, Hong Kong, Hong Kong, Peoples R China
基金
美国国家科学基金会;
关键词
Computer vision; Construction safety; Guardrail detection; Convolutional neural networks; Transfer learning; VGG-16; CONSTRUCTION WORKERS; OBJECT DETECTION; RECOGNITION; SYSTEM; MANAGEMENT; RESOURCES; TRACKING; POSTURE;
D O I
10.1016/j.autcon.2018.01.003
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Safety has been a concern for the construction industry for decades. Unsafe conditions and behaviors are considered as the major causes of construction accidents. The current safety inspection of conditions and behaviors heavily rely on human efforts which are limited onsite. To improve the safety performance of the industry, a more efficient approach to identify the unsafe conditions on site is required to supplement the current manual inspection practice. A promising way to supplement the current manual safety inspection is automated and intelligent monitoring/inspection through information and sensing technologies, including localization techniques, environment monitoring, image processing and etc. To assess the potential benefits of contemporary technologies for onsite safety inspection, the authors focused on real-time guardrail detection, as unprotected edges are the ones cause for workers falling from heights. In this paper, the authors developed a safety guardrail detection model based on convolutional neural network (CNN). An augmented data set is generated with the addition of background image to guardrail 3D models and used as training set. Transfer learning is utilized and the Visual Geometry Group architecture with 16 layers (VGG-16) model is adopted to construct the basic features extraction for the neural network. In the CNN implementation, 4000 augmented images were used to train the proposed model, while another 2000 images collected from real construction jobsites and 2000 images from Google were used to validate the proposed model. The proposed CNN-based guardrail detection model obtained a high accuracy of 96.5%. In addition, this study indicates that the synthetic images generated by augment technology can be used to create a large training dataset, and CNN-based image detection algorithm is a promising approach in construction jobsite safety monitoring.
引用
收藏
页码:58 / 70
页数:13
相关论文
共 45 条
[1]  
[Anonymous], 2012, Technical Report
[2]  
[Anonymous], P AS C COMP VIS ACCV, DOI DOI 10.1007/978-3-642-19315-6_16
[3]  
Brilakis I, 2011, ADV ENG INFORM, V25, DOI [10.1016/j.ael.2011.01.003, DOI 10.1016/J.AEL.2011.01.003]
[4]   A survey of free-form object representation and recognition techniques [J].
Campbell, RJ ;
Flynn, PJ .
COMPUTER VISION AND IMAGE UNDERSTANDING, 2001, 81 (02) :166-210
[5]   A proactive system for real-time safety management in construction sites [J].
Carbonari, Alessandro ;
Giretti, Alberto ;
Naticchia, Berardo .
AUTOMATION IN CONSTRUCTION, 2011, 20 (06) :686-698
[6]   A survey on object detection in optical remote sensing images [J].
Cheng, Gong ;
Han, Junwei .
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING, 2016, 117 :11-28
[7]   Real-time resource location data collection and visualization technology for construction safety and activity monitoring applications [J].
Cheng, Tao ;
Teizer, Jochen .
AUTOMATION IN CONSTRUCTION, 2013, 34 :3-15
[8]   Automated task-level activity analysis through fusion of real time location sensors and worker's thoracic posture data [J].
Cheng, Tao ;
Teizer, Jochen ;
Migliaccio, Giovanni C. ;
Gatti, Umberto C. .
AUTOMATION IN CONSTRUCTION, 2013, 29 :24-39
[9]  
Chow Y. Y., 2017, CASE STUDIES ACCIDEN
[10]  
Du S., 2011, COMPUTER RES DEV ICC, V4, P25, DOI DOI 10.1109/ICCRD.2011.5763846