SODA: Site Object Detection dAtaset for Deep Learning in Construction

Published:

Together with Dr. Yichuan Deng’s group from South China University of Technology, we developed a large-scale dataset called SODA (Site Object Detection dAtaset) with about 20,000 images, for deep learning based site object detection. Images in SODA are taken from different view angles and ambient lighting conditions, and covers various site objects such as workers, equipments, materials, etc. Experiements show that the proposed data could improve the performance of deep learning models such as YOLO significantly. Details and development of SODA could be found in our paper, and the dataset could be accessed here for future investigations.

We name the dataset SODA in the hope that it will bring hapiness to researchers and practitioners in the construction area:)

Leave a Comment