Abstract
In modern construction, the construction sites are congested, busy, and full of obstacles. The environment in the construction site is dynamic. In recent times, deep learning-based models remain the main tools used for image classification. However, the performance of the deep learning models in construction site-related image classification is not convincing due to the dynamic nature and busyness of the construction sites. The availability of construction site-related image datasets also remains an obstacle in achieving the best performance from the deep learning models. Data augmentation is a technique used to apply random but realistic transformation to the images. Data augmentation will not only help to diversify the image dataset but also assist in increasing the size of the dataset. This study used a state-of-the-art YOLOv4 deep learning model and implemented data augmentation techniques like gamma transformation to control the intensity of light and mimic sunny, cloudy, day, and night situations in the construction site images. The other data augmentation technique used is Gaussian blur to minimize the details in the images, and salt-and-pepper noise to degrade the quality of construction site images. The model is trained and tested on Alberta Construction Image Dataset (ACID) and construction workers hand signal image datasets, with and without the implementation of data augmentation. The performance of the model is evaluated based on the test dataset while keeping all the parameters of the model the same. It is observed that the model trained on the augmented dataset performed better than the model trained on the non-augmented dataset by 4%.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bochkovskiy A, Wang C, Liao HM (2020) YOLOv4: optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934
Boncelet C (2009) Image noise models. In: The essential guide to image processing. Academic Press, pp 143–167
Bull D (2014) Digital picture formats and representations. Communicating Pictures 99–132
Davis P, Aziz F, Newaz MT, Sher W, Simon L (2021) The classification of construction waste material using a deep convolutional neural network. Autom Constr, Elsevier 122:103481
Fang Q, Li H, Luo X, Ding L, Luo H, Rose TM, An W (2018) Detecting non-hardhat-use by a deep learning method from far-field surveillance videos. Autom Constr, Elsevier 85:1–9
Gu S, Pednekar M, Slater R (2019) Improve image classification using data augmentation and neural networks. SMU Data Sci Rev 2(2):1–43
Hernandez-Garcia A, Konig P (2018) Further advantages of data augmentation on convolutional neural networks. Springer Nature Switzerland, pp 95–103
Hernández-GarcÃa A, König P (2018) Further advantages of data augmentation on convolutional neural networks. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), pp 95–103.
Hou X, Zhang Y, Hou J (2021) Application of YOLO V2 in construction vehicle detection. Springer International Publishing, Lecture Notes on Data Engineering and Communications Technologies
Hu J, Geo X, Wu H, Gao S (2019) Detection of workers without the helments in videos based on YOLO V3. In: 12th International congress on image and signal processing, biomedical engineering and informatics, pp 1553–1560
Huang Z, Wang J, Fu X, Yu T, Guo Y, Wang R (2020) DC-SPP-YOLO: dense connection and spatial pyramid pooling based YOLO for object detection. Inf Sci 522:241–258
Lei C, Hu B, Wang D, Zhang S, Chen Z (2019, October) A preliminary study on data augmentation of deep learning for image classification. In: Proceedings of the 11th Asia-pacific symposium on internetware, pp 1–6
Liu S, Qi L, Qin H, Shi J, Jia J (2018) Path Aggregation network for instance segmentation. In: Proceedings of the IEEE computer society conference on computer vision and pattern recognition, pp 8759–8768
Mikołajczyk A, Grochowski M (2018) Data augmentation for improving deep learning in image classification problem. In: International interdisciplinary PhD workshop (IIPhDW). IEEE, pp 117–122
Misra S, Wu Y (2020) Machine learning assisted segmentation of scanning electron microscopy images of organic-rich shales with feature extraction and feature ranking. Mach Learn Subsurf Charact 289
Perez L, Wang J (2017) The effectiveness of data augmentation in image classification using deep learning. arXiv preprint arXiv:1712.04621
Shorten C, Khoshgoftaar TM (2019) A survey on image data augmentation for deep learning. J Big Data, Springer Int Publishing 6(1):1–48
Walawalkar D, Shen Z, Liu Z, Savvides M (2020) Attentive cutmix: an enhanced data augmentation approach for deep learning based image classification. In: ICASSP, IEEE international conference on acoustics, speech and signal processing - proceedings, pp 3642–3646
Xiao B, Kang S-C (2021a) Vision-based method integrating deep learning detection for tracking multiple construction machines. J Comput Civ Eng 35(2):04020071
Xiao B, Kang S-C (2021b) Development of an image data set of construction machines for deep learning object detection. J Comput Civ Eng 35(2):05020005
Yin X, Chen Y, Bouferguene A, Zaman H, Al-Hussein M, Kurach L (2020) A deep learning-based framework for an automated defect detection system for sewer pipes. Autom Constr 109:102967
Yu Y, Li H, Umer W, Dong C, Yang X, Skitmore M, Wong AYL (2019) Automatic biomechanical workload estimation for construction workers by computer vision and smart insoles. J Comput Civ Eng 33(3):1–13
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 Canadian Society for Civil Engineering
About this paper
Cite this paper
Asif, M., Liu, S., Ali, G.M., Bouferguene, A., Al-Hussein, M. (2023). The Effectiveness of Data Augmentation in Construction Site-Related Image Classification. In: Gupta, R., et al. Proceedings of the Canadian Society of Civil Engineering Annual Conference 2022. CSCE 2022. Lecture Notes in Civil Engineering, vol 363. Springer, Cham. https://doi.org/10.1007/978-3-031-34593-7_16
Download citation
DOI: https://doi.org/10.1007/978-3-031-34593-7_16
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-34592-0
Online ISBN: 978-3-031-34593-7
eBook Packages: EngineeringEngineering (R0)