Abstract
CV based on deep learning is popular for detecting damages of infrastructures, the crack image segmentation by UNet model was investigated in this work. 120 images with resolutions either 4928x3264 or 5152x3864 pixels were selected as dataset which contains damage images of steel structures. Crack and non-crack images were selected for dataset at a ratio of 1:1. Among of them, 110 images were used for training dataset and the other 10 images were selected for model testing. This research compared the results from different models trained by BCE, MSE and L1 losses. The results have validated that the BCE loss based model demonstrated the best performance with the mIoU (76.66%) higher than MSE-loss-model (4.4%) and L1-loss-model (11.61%), respectively.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
