Automatic identification algorithm for insulation board bonding status based on improved YOLOv7

• Published in: 2024 IEEE International Conference on Signal, Information and Data Processing (ICSIDP) pp. 1 - 6
• Main Authors: Li, Yuhan, Wang, Jian, Jiang, Zihang, Liu, Mian, Gong, Junbo, Lan, Tian
• Format: Conference Proceeding
• Language: English
• Published: IEEE 22.11.2024
• Subjects: Accuracy; Attention mechanisms; Bonding; Bonding status identification; Buildings; Data processing; Exterior wall insulation board; Ground penetrating radar; Ground-penetrating radar (GPR); Insulation; Manuals; Radar detection; Radar imaging; YOLOv7
• DOI: 10.1109/ICSIDP62679.2024.10868355

At present, for the detection of the bonding status of the building exterior wall insulation layer mainly relies on manual interpretation of radar images, which is not only time-consuming and prone to errors; at the same time, the background of the ground-penetrating radar image is relatively complex, so how to accurately determine whether it is the normal bonding state or the top debonding state or the subgrade debonding state between the insulation board and the bonding agent or between the wall and the bonding agent is still a difficult problem. For the detection of the three different states of normal bonding, top debonding and subgrade debonding, the YOLOv7 algorithm is used to achieve automatic identification; on this basis, in order to improve the phenomenon of multi-inspection and misdetection, the BiFormer attention mechanism is introduced after the ELAN module of YOLOv7, and the improvement of the YOLOv7's mAP@0.5:0.95 can reach 0.7788, and the phenomenon of multi-detection and misdetection is greatly reduced, which verifies the feasibility and accuracy of the improved model, and significantly improves the accuracy and efficiency of the model for the identification of defects in the bonding status of the insulation board.