2023 Best Paper Award

The Editorial Board of the ASCE Journal of Construction Engineering and Management is pleased to announce the recipients of the 2023 Best Paper Award: Bo Xiao, Yiheng Wang, and Shih-Chung Kang are coauthors of an article entitled “Deep Learning Image Captioning in Construction Management: A Feasibility Study,” which appeared in the July 2022 issue of the Journal of Construction Engineering and Management, https://doi.org/10.1061/(ASCE)CO.1943-7862.0002297.

Construction processes are often documented with a plethora of images, and this makes it highly desirable to have information technology processes to categorize, retrieve, and analyze those images. Descriptive captions are amenable to efficient text-based search techniques but are time consuming to develop manually, especially given the volume of images needed to properly document a construction process.

The authors present their systematic approach to automating deep learning for image captioning in a construction management context. Several steps are presented including developing a linguistic schema to standardize the captions, creating a data set of manually captioned images for model training and evaluation, selection of six deep-learning methods for image captioning, and use of eight metrics for experimental evaluation. Each of the six methods were applied to a database of 4,000 manually captioned images and the results compared according to the eight metrics.

A rich set of conclusions are offered. A method that combined two advanced deep learning processes had the best performance according to all eight measures. This outcome was compared with successful automatic captioning efforts outside of construction management and found to be on par; this gives encouragement that such captioning will be feasible for construction management. The authors also noted that certain image analysis tasks would be better executed with previously available methods.

This article is compelling for several reasons. The topic is one that has the potential to provide substantial benefits for construction engineering and management. The authors have clearly explained an involved methodology that produced results that were validated by comparison of eight metrics. The conclusions are presented in a nuanced manner that includes the strong points and limitations of the methods studied and useful suggestions for future development.

This paper was selected from a field of 187 technical papers, 25 case studies, 2 technical notes, and 18 state-of-the-art reviews that were published from July 2022 through June 2023 in the Journal of Construction Engineering and Management. For this awards cycle, the Awards Committee included the following editorial board members: Dr. Young Hoon Kwak, Dr. Young Bai, Dr. Mehmet E. Ozbek, Dr. Carolyn M. Clevenger, Dr. Caldas Carlos, Dr. SangHyun Lee, Dr. Yilong Han, Dr. Reza Akhavian, and Dr. Ioannis Brilakis, with Dr. Charles T. Jahren serving as the chair. Special thanks are due to Dr. Daniel Tran, Min Liu, and Dr. Fei Dai for providing additional reviews for this and other ASCE awards. Taryn Dollings provided administrative support as an ASCE Journals staff member. Thanks to the other editorial board members, authors, reviewers, and readers who help to make this journal a success and this award meaningful.